def add_metric_plot(self,
plot_collection,
dataset_name,
metric,
fig_title="",
plot_title="",
metric_units=""):
""" Adds a metric plot to a plot collection.
Args:
plot_collection: a PlotCollection containing plots.
dataset_name: a string representing the name of the dataset being evaluated.
metric: an evo.core.metric object with statistics and information.
fig_title: a string representing the title of the figure. Must be unique in the plot_collection.
plot_title: a string representing the title of the plot.
metric_units: a string representing the units of the metric being plotted.
"""
fig = plt.figure(figsize=(8, 8))
stats = metric.get_all_statistics()
plot.error_array(fig,
metric.error,
statistics=stats,
name=plot_title,
title=plot_title,
xlabel="Keyframe index [-]",
ylabel=plot_title + " " + metric_units)
plot_collection.add_figure(fig_title, fig)
def plot(args, result, traj_ref, traj_est):
from evo.tools import plot
from evo.tools.settings import SETTINGS
import matplotlib.pyplot as plt
import numpy as np
logger.debug(SEP)
logger.debug("Plotting results... ")
plot_mode = plot.PlotMode(args.plot_mode)
# Plot the raw metric values.
fig1 = plt.figure(figsize=SETTINGS.plot_figsize)
if "seconds_from_start" in result.np_arrays:
seconds_from_start = result.np_arrays["seconds_from_start"]
else:
seconds_from_start = None
plot.error_array(
fig1, result.np_arrays["error_array"], x_array=seconds_from_start,
statistics={
s: result.stats[s]
for s in SETTINGS.plot_statistics if s not in ("min", "max")
}, name=result.info["label"], title=result.info["title"],
xlabel="$t$ (s)" if seconds_from_start else "index")
# Plot the values color-mapped onto the trajectory.
fig2 = plt.figure(figsize=SETTINGS.plot_figsize)
ax = plot.prepare_axis(fig2, plot_mode)
plot.traj(ax, plot_mode, traj_ref, style=SETTINGS.plot_reference_linestyle,
color=SETTINGS.plot_reference_color, label='reference',
alpha=SETTINGS.plot_reference_alpha)
if args.plot_colormap_min is None:
args.plot_colormap_min = result.stats["min"]
if args.plot_colormap_max is None:
args.plot_colormap_max = result.stats["max"]
if args.plot_colormap_max_percentile is not None:
args.plot_colormap_max = np.percentile(
result.np_arrays["error_array"], args.plot_colormap_max_percentile)
plot.traj_colormap(ax, traj_est, result.np_arrays["error_array"],
plot_mode, min_map=args.plot_colormap_min,
max_map=args.plot_colormap_max,
title="Error mapped onto trajectory")
fig2.axes.append(ax)
plot_collection = plot.PlotCollection(result.info["title"])
plot_collection.add_figure("raw", fig1)
plot_collection.add_figure("map", fig2)
if args.plot:
plot_collection.show()
if args.save_plot:
plot_collection.export(args.save_plot,
confirm_overwrite=not args.no_warnings)
if args.serialize_plot:
logger.debug(SEP)
plot_collection.serialize(args.serialize_plot,
confirm_overwrite=not args.no_warnings)
def plot_metric(metric, plot_title="", figsize=(8, 8)):
""" Adds a metric plot to a plot collection.
Args:
plot_collection: a PlotCollection containing plots.
metric: an evo.core.metric object with statistics and information.
plot_title: a string representing the title of the plot.
figsize: a 2-tuple representing the figure size.
Returns:
A plt figure.
"""
fig = plt.figure(figsize=figsize)
stats = metric.get_all_statistics()
plot.error_array(fig,
metric.error,
statistics=stats,
title=plot_title,
xlabel="Keyframe index [-]",
ylabel=plot_title + " " + metric.unit.value)
return fig
print("calculating APE")
data = (traj_ref, traj_est)
ape_metric = metrics.APE(metrics.PoseRelation.translation_part)
ape_metric.process_data(data)
ape_statistics = ape_metric.get_all_statistics()
print("mean:", ape_statistics["mean"])
print("loading plot modules")
from evo.tools import plot
import matplotlib.pyplot as plt
print("plotting")
plot_collection = plot.PlotCollection("Example")
# metric values
fig_1 = plt.figure(figsize=(8, 8))
plot.error_array(fig_1, ape_metric.error, statistics=ape_statistics,
name="APE", title=str(ape_metric))
plot_collection.add_figure("raw", fig_1)
# trajectory colormapped with error
fig_2 = plt.figure(figsize=(8, 8))
plot_mode = plot.PlotMode.xy
ax = plot.prepare_axis(fig_2, plot_mode)
plot.traj(ax, plot_mode, traj_ref, '--', 'gray', 'reference')
plot.traj_colormap(ax, traj_est, ape_metric.error, plot_mode,
min_map=ape_statistics["min"], max_map=ape_statistics["max"],
title="APE mapped onto trajectory")
plot_collection.add_figure("traj (error)", fig_2)
# trajectory colormapped with speed
fig_3 = plt.figure(figsize=(8, 8))
plot_mode = plot.PlotMode.xy
def main_ape(traj_ref,
traj_est,
pose_relation,
align=True,
correct_scale=False,
ref_name="",
est_name="",
show_plot=False,
save_plot=None,
plot_mode=None,
save_results=None,
no_warnings=False,
serialize_plot=None):
from evo.algorithms import metrics
from evo.algorithms import trajectory
from evo.tools import file_interface
from evo.tools.settings import SETTINGS
only_scale = correct_scale and not align
if align or correct_scale:
logging.debug(SEP)
if only_scale:
logging.debug("correcting scale...")
else:
logging.debug("aligning using Umeyama's method..." + (
" (with scale correction)" if correct_scale else ""))
traj_est = trajectory.align_trajectory(traj_est, traj_ref,
correct_scale, only_scale)
logging.debug(SEP)
# calculate APE
data = (traj_ref, traj_est)
ape_metric = metrics.APE(pose_relation)
ape_metric.process_data(data)
ape_statistics = ape_metric.get_all_statistics()
title = str(ape_metric)
if align and not correct_scale:
title += "\n(with SE(3) Umeyama alignment)"
elif align and correct_scale:
title += "\n(with Sim(3) Umeyama alignment)"
elif only_scale:
title += "\n(scale corrected)"
else:
title += "\n(not aligned)"
logging.debug(SEP)
res_str = ""
for name, val in sorted(ape_statistics.items()):
res_str += "{:>10}".format(name) + "\t" + "{0:.6f}".format(val) + "\n"
logging.info("\nstatistics of " + title + ":\n\n" + res_str)
if show_plot or save_plot or save_results or serialize_plot:
if isinstance(traj_est, trajectory.PoseTrajectory3D):
seconds_from_start = [
t - traj_est.timestamps[0] for t in traj_est.timestamps
]
else:
seconds_from_start = None
if show_plot or save_plot or serialize_plot:
from evo.tools import plot
import matplotlib.pyplot as plt
logging.debug(SEP)
logging.debug("plotting results... ")
fig1 = plt.figure(figsize=(SETTINGS.plot_figsize[0],
SETTINGS.plot_figsize[1]))
# metric values
plot.error_array(
fig1,
ape_metric.error,
x_array=seconds_from_start,
statistics=ape_statistics,
name="APE" + (" (" + ape_metric.unit.value + ")")
if ape_metric.unit else "",
title=title,
xlabel="$t$ (s)" if seconds_from_start else "index")
# info text
if SETTINGS.plot_info_text and est_name and ref_name:
ax = fig1.gca()
ax.text(0,
-0.12,
"estimate: " + est_name + "\nreference: " + ref_name,
transform=ax.transAxes,
fontsize=8,
color="gray")
# trajectory colormapped
fig2 = plt.figure(figsize=(SETTINGS.plot_figsize[0],
SETTINGS.plot_figsize[1]))
plot_mode = plot_mode if plot_mode is not None else plot.PlotMode.xyz
ax = plot.prepare_axis(fig2, plot_mode)
plot.traj(ax,
plot_mode,
traj_ref,
'--',
'gray',
'reference',
alpha=0.0 if SETTINGS.plot_hideref else 1.0)
plot.traj_colormap(ax,
traj_est,
ape_metric.error,
plot_mode,
min_map=ape_statistics["min"],
max_map=ape_statistics["max"],
title="APE mapped onto trajectory")
fig2.axes.append(ax)
plot_collection = plot.PlotCollection(title)
plot_collection.add_figure("raw", fig1)
plot_collection.add_figure("map", fig2)
if show_plot:
plot_collection.show()
if save_plot:
plot_collection.export(save_plot,
confirm_overwrite=not no_warnings)
if serialize_plot:
logging.debug(SEP)
plot_collection.serialize(serialize_plot,
confirm_overwrite=not no_warnings)
if save_results:
logging.debug(SEP)
file_interface.save_res_file(save_results,
ape_metric,
ape_statistics,
title,
ref_name,
est_name,
seconds_from_start,
traj_ref,
traj_est,
confirm_overwrite=not no_warnings)
return ape_statistics, ape_metric.error
def get_and_save_results_from_folder(folder_with_predicted_poses,category):
global args
global kitti_eval_tool
global folder_with_gt_poses
global output_folder
global t
global results
values_for_excel = []
columns_for_excel = []
type_of_statistics = 'mean'
for filename in sorted(os.listdir(folder_with_predicted_poses)):
if not(os.path.exists(os.path.join(folder_with_gt_poses, filename))):
print("file with gt poses doesn't exist for "+filename)
continue
if filename.find('.txt') == -1:
continue
seq_results = {}
seq_results['name_seq'] = filename[:filename.rfind('.')]
seq_results['category'] = category
folder_name = seq_results['category']
seq_results['metrics'] = {}
seq_results['lost'] = False
os.makedirs(os.path.join(output_folder, folder_name), exist_ok=True)
output_folder_seq = os.path.join(output_folder, folder_name, filename[:filename.rfind('.')])
os.makedirs(output_folder_seq, exist_ok=True)
if os.path.isfile(os.path.join(output_folder, folder_name,"results.txt")):
file_results_txt = open(os.path.join(output_folder, folder_name,"results.txt"), "a")
else:
file_results_txt = open(os.path.join(output_folder, folder_name,"results.txt"), "w")
file_results_txt.write("translation_error(%) rotation_error(deg/m) ATE(m) APE_translation_error_median(m) APE_rotation_error_median(deg) dst_to_trgt\n")
# -------------------------------------Getting results---------------------------------------------------
if args.gt_format == 'kitti':
traj_ref = file_interface.read_kitti_poses_file(os.path.join(folder_with_gt_poses, filename))
if args.gt_format == 'tum':
traj_ref = file_interface.read_tum_trajectory_file(os.path.join(folder_with_gt_poses, filename))
seq_results["length_of_ref_traj"] = traj_ref.path_length
end_time_gt = traj_ref.get_infos()["t_end (s)"]
if args.gt_format == 'euroc':
traj_ref = file_interface.read_euroc_csv_trajectory(os.path.join(folder_with_gt_poses, filename))
if args.result_format == 'kitti':
traj_est = file_interface.read_kitti_poses_file(os.path.join(folder_with_predicted_poses, filename))
if args.result_format == 'tum':
traj_est = file_interface.read_tum_trajectory_file(os.path.join(folder_with_predicted_poses, filename))
seq_results["length_of_estimated_traj"] = traj_est.path_length
if args.result_format == 'euroc':
traj_est = file_interface.read_euroc_csv_trajectory(os.path.join(folder_with_predicted_poses, filename))
if args.result_format == 'tum' and args.gt_format == 'tum':
seq_results["num_gt_poses"] = traj_ref.num_poses
seq_results["num_predicted_poses"] = traj_est.num_poses
traj_ref, traj_est = sync.associate_trajectories(traj_ref, traj_est, args.max_diff)
end_time_est = traj_est.get_infos()["t_end (s)"]
if (abs(end_time_est - end_time_gt) > 0.2) or (traj_est.get_infos()["t_start (s)"] > 0.2):
print('LOST in track '+filename[:filename.rfind('.')])
seq_results['lost'] = True
results.append(seq_results)
t.update(1)
continue
if args.alignment != None:
traj_est = trajectory.align_trajectory(traj_est, traj_ref, correct_scale=args.alignment.find("scale") != -1, correct_only_scale=args.alignment=="scale")
trajectory.align_trajectory_origin(traj_est, traj_ref)
data = (traj_ref, traj_est)
ape_metric_translation = metrics.APE(metrics.PoseRelation.translation_part)
ape_metric_rotation = metrics.APE(metrics.PoseRelation.rotation_angle_deg)
ape_metric_translation.process_data(data)
ape_metric_rotation.process_data(data)
ape_translation_statistics = ape_metric_translation.get_all_statistics()
ape_rotation_statistics = ape_metric_rotation.get_all_statistics()
ape_translation_statistics_plot = copy.deepcopy(ape_translation_statistics)
ape_rotation_statistics_plot = copy.deepcopy(ape_rotation_statistics)
ape_translation_statistics_plot.pop('sse')
ape_translation_statistics_plot.pop('std')
ape_translation_statistics_plot.pop('min')
ape_translation_statistics_plot.pop('max')
ape_rotation_statistics_plot.pop('sse')
ape_rotation_statistics_plot.pop('std')
ape_rotation_statistics_plot.pop('min')
ape_rotation_statistics_plot.pop('max')
kitti_trans_err, kitti_rot_err, ate = kitti_eval_tool.eval(traj_ref.poses_se3,
traj_est.poses_se3,
alignment=None)
#---------------------------------adding results to variable seq_results for excel -----------------------------
seq_results['metrics']['dist_to_trgt'] = traj_est.get_infos()['pos_end (m)'] - traj_ref.get_infos()['pos_end (m)']
seq_results['metrics']['dist_to_trgt'] = np.sum(np.array(seq_results['metrics']['dist_to_trgt'])**2)**0.5
seq_results['metrics']["Kitti trans err (%)"] = kitti_trans_err
seq_results['metrics']["Kitti rot err (deg/m)"] = kitti_rot_err
seq_results['metrics']["ATE (m)"] = ate
seq_results['metrics']["APE(trans err) median (m)"] = ape_translation_statistics["median"]
seq_results['metrics']["APE(rot err) median (deg)"] = ape_rotation_statistics["median"]
#--------------------------------------------------------------------------------------------------------
#-------------------------------------------------------------------------------------------------------
# --------------------------------printing results into console----------------------------------------------
print('Results for "'+filename+'":')
print('Kitti average translational error (%): {:.7f}'.format(kitti_trans_err))
print('Kitti average rotational error (deg/m): {:.7f}'.format(kitti_rot_err))
print('ATE (m): {:.7f}'.format(ate))
print('APE(translation error) median (m): {:.7f}'.format(ape_translation_statistics["median"]))
print('APE(rotation error) median (deg): {:.7f}'.format(ape_rotation_statistics["median"]))
print('distance to target on the last frame: {:.7f}'.format(seq_results['metrics']['dist_to_trgt']))
#------------------------------------------------------------------------------------------------------------
#---------------------------------Saving results into overall results text file------------------------------
file_results_txt.write('{:<24} '.format(filename[:filename.rfind('.')]))
file_results_txt.write('{:>7.4f} '.format(kitti_trans_err))
file_results_txt.write('{:>7.4f} '.format(kitti_rot_err))
file_results_txt.write('{:>7.4f} '.format(ate))
file_results_txt.write('{:>7.4f} '.format(ape_translation_statistics["median"]))
file_results_txt.write('{:>7.4f} '.format(ape_rotation_statistics["median"]))
file_results_txt.write('{:>7.4f}\n'.format(seq_results['metrics']['dist_to_trgt']))
#------------------------------------------------------------------------------------------------------------
# --------------------------------Saving metrics to text file for one track----------------------------------
txt_filename = filename[:filename.rfind('.')]+"_metrics.txt"
with open(os.path.join(output_folder_seq, txt_filename), "w") as txt_file:
txt_file.write('Kitti average translational error (%): {:.7f}\n'.format(kitti_trans_err))
txt_file.write('Kitti average rotational error (deg/m): {:.7f}\n'.format(kitti_rot_err))
txt_file.write('ATE (m): {:.7f}\n'.format(ate))
txt_file.write('APE(translation error) median (m): {:.7f}\n'.format(ape_translation_statistics["median"]))
txt_file.write('APE(rotation error) median (deg): {:.7f}\n'.format(ape_rotation_statistics["median"]))
txt_file.write('Distance to target on the last frame: {:.7f}\n'.format(seq_results['metrics']['dist_to_trgt']))
#---------------------------------------------------------------------------------------------------------
# ---------------------------------Saving values of errors for each frame to text file------------------------
# ------------------------------------------for translation errors----------------------------------------
txt_filename = filename[:filename.rfind('.')]+"_APE(translation)_errors.txt"
output_folder_seq_translation = os.path.join(output_folder_seq,"translation")
output_folder_seq_rotation = os.path.join(output_folder_seq,"rotation")
os.makedirs(output_folder_seq_translation, exist_ok=True)
os.makedirs(output_folder_seq_rotation, exist_ok=True)
with open(os.path.join(output_folder_seq_translation, txt_filename), "w") as txt_file:
for error in ape_metric_translation.error:
txt_file.write('{:.10f}\n'.format(error))
# -----------------------------------------for rotation degree errors--------------------------------------
txt_filename = filename[:filename.rfind('.')]+"_APE(rotation_deg)_errors.txt"
with open(os.path.join(output_folder_seq_rotation, txt_filename), "w") as txt_file:
for error in ape_metric_rotation.error:
txt_file.write('{:.10f}\n'.format(error))
#----------------------------------------------------------------------------------------------------------
# ---------------------------------------Saving plot of errors of each frame------------------------------
# ------------------------------------------for translation errors----------------------------------------
plot_collection = plot.PlotCollection("Example")
fig_1 = plt.figure(figsize=(8, 8))
plot.error_array(fig_1, ape_metric_translation.error,
name="APE", title=str(ape_metric_translation), xlabel="Index of frame", ylabel='Error')
plot_collection.add_figure("raw", fig_1)
plot_filename = filename[:filename.rfind('.')]+"_APE(translation)_errors.png"
plt.savefig(os.path.join(output_folder_seq_translation, plot_filename))
plt.close(fig_1)
# -----------------------------------------for rotation degree errors--------------------------------------
plot_collection = plot.PlotCollection("Example")
fig_1 = plt.figure(figsize=(8, 8))
plot.error_array(fig_1, ape_metric_rotation.error,
name="APE", title=str(ape_metric_rotation), xlabel="Index of frame", ylabel='Error')
plot_collection.add_figure("raw", fig_1)
plot_filename = filename[:filename.rfind('.')]+"_APE(rotation)_errors.png"
plt.savefig(os.path.join(output_folder_seq_rotation,plot_filename))
plt.close(fig_1)
#-----------------------------------------------------------------------------------------------------------
# -----------------------------------------Saving trajectory plot-------------------------------------------
# ------------------------------------------for translation errors----------------------------------------
fig_2 = plt.figure(figsize=(8, 8))
ax = plot.prepare_axis(fig_2, plot_mode)
plot.traj(ax, plot_mode, traj_ref, '--', 'gray', 'reference')
plot.traj_colormap( ax, traj_est, ape_metric_translation.error, plot_mode,
min_map=ape_translation_statistics["min"],
max_map=ape_translation_statistics["max"], title="APE translation mapped onto trajectory")
plot_collection.add_figure("traj (error)", fig_2)
plot_filename = filename[:filename.rfind('.')]+"_APE(translation)_map.png"
plt.savefig(os.path.join(output_folder_seq_translation,plot_filename))
plt.close(fig_2)
# -----------------------------------------for rotation degree errors--------------------------------------
fig_2 = plt.figure(figsize=(8, 8))
ax = plot.prepare_axis(fig_2, plot_mode)
plot.traj(ax, plot_mode, traj_ref, '--', 'gray', 'reference')
plot.traj_colormap( ax, traj_est, ape_metric_rotation.error, plot_mode,
min_map=ape_rotation_statistics["min"],
max_map=ape_rotation_statistics["max"], title="APE rotation mapped onto trajectory")
plot_collection.add_figure("traj (error)", fig_2)
plot_filename = filename[:filename.rfind('.')]+"_APE(rotation)_map.png"
plt.savefig(os.path.join(output_folder_seq_rotation,plot_filename))
plt.close(fig_2)
#-----------------------------------------------------------------------------------------------------------
print()
active_worksheet = wb['sheet1']
thin = Side(border_style="thin", color="000000")
thick = Side(border_style="thick", color="000000")
medium = Side(border_style="medium", color="000000")
font_header = Font(name='Arial',
size=10,
bold=True,
italic=False,
vertAlign=None,
underline='none',
strike=False,
color='FF000000')
font_values = Font(name='Arial',
size=10,
bold=False,
italic=False,
vertAlign=None,
underline='none',
strike=False,
color='FF000000')
active_worksheet.row_dimensions[2].height = 35
file_results_txt.close()
results.append(seq_results)
t.update(1)
ape_metric = metrics.APE(metrics.PoseRelation.translation_part)
ape_metric.process_data(data)
ape_statistics = ape_metric.get_all_statistics()
print("mean:", ape_statistics["mean"])
print("loading plot modules")
from evo.tools import plot
import matplotlib.pyplot as plt
print("plotting")
plot_collection = plot.PlotCollection("Example")
# metric values
fig_1 = plt.figure(figsize=(8, 8))
plot.error_array(fig_1,
ape_metric.error,
statistics=ape_statistics,
name="APE",
title=str(ape_metric))
plot_collection.add_figure("raw", fig_1)
# trajectory colormapped with error
fig_2 = plt.figure(figsize=(8, 8))
plot_mode = plot.PlotMode.xy
ax = plot.prepare_axis(fig_2, plot_mode)
plot.traj(ax, plot_mode, traj_ref, '--', 'gray', 'reference')
plot.traj_colormap(ax,
traj_est,
ape_metric.error,
plot_mode,
min_map=ape_statistics["min"],
max_map=ape_statistics["max"],
def plot_result(args: argparse.Namespace,
result: Result,
traj_ref: PosePath3D,
traj_est: PosePath3D,
traj_ref_full: typing.Optional[PosePath3D] = None) -> None:
from evo.tools import plot
from evo.tools.settings import SETTINGS
import matplotlib.pyplot as plt
import numpy as np
logger.debug(SEP)
logger.debug("Plotting results... ")
plot_mode = plot.PlotMode(args.plot_mode)
# Plot the raw metric values.
fig1 = plt.figure(figsize=SETTINGS.plot_figsize)
if "seconds_from_start" in result.np_arrays:
seconds_from_start = result.np_arrays["seconds_from_start"]
else:
seconds_from_start = None
plot.error_array(
fig1.gca(),
result.np_arrays["error_array"],
x_array=seconds_from_start,
statistics={
s: result.stats[s]
for s in SETTINGS.plot_statistics if s not in ("min", "max")
},
name=result.info["label"],
title=result.info["title"],
xlabel="$t$ (s)" if seconds_from_start is not None else "index")
# Plot the values color-mapped onto the trajectory.
fig2 = plt.figure(figsize=SETTINGS.plot_figsize)
ax = plot.prepare_axis(fig2, plot_mode)
if args.ros_map_yaml:
plot.ros_map(ax, args.ros_map_yaml, plot_mode)
plot.traj(ax,
plot_mode,
traj_ref_full if traj_ref_full else traj_ref,
style=SETTINGS.plot_reference_linestyle,
color=SETTINGS.plot_reference_color,
label='reference',
alpha=SETTINGS.plot_reference_alpha)
plot.draw_coordinate_axes(ax, traj_ref, plot_mode,
SETTINGS.plot_axis_marker_scale)
if args.plot_colormap_min is None:
args.plot_colormap_min = result.stats["min"]
if args.plot_colormap_max is None:
args.plot_colormap_max = result.stats["max"]
if args.plot_colormap_max_percentile is not None:
args.plot_colormap_max = np.percentile(
result.np_arrays["error_array"], args.plot_colormap_max_percentile)
plot.traj_colormap(ax,
traj_est,
result.np_arrays["error_array"],
plot_mode,
min_map=args.plot_colormap_min,
max_map=args.plot_colormap_max,
title="Error mapped onto trajectory")
plot.draw_coordinate_axes(ax, traj_est, plot_mode,
SETTINGS.plot_axis_marker_scale)
if SETTINGS.plot_pose_correspondences:
plot.draw_correspondence_edges(
ax,
traj_est,
traj_ref,
plot_mode,
style=SETTINGS.plot_pose_correspondences_linestyle,
color=SETTINGS.plot_reference_color,
alpha=SETTINGS.plot_reference_alpha)
fig2.axes.append(ax)
plot_collection = plot.PlotCollection(result.info["title"])
plot_collection.add_figure("raw", fig1)
plot_collection.add_figure("map", fig2)
if args.plot:
plot_collection.show()
if args.save_plot:
plot_collection.export(args.save_plot,
confirm_overwrite=not args.no_warnings)
if args.serialize_plot:
logger.debug(SEP)
plot_collection.serialize(args.serialize_plot,
confirm_overwrite=not args.no_warnings)
def main_rpe_for_each(traj_ref,
traj_est,
pose_relation,
mode,
bins,
rel_tols,
align=False,
correct_scale=False,
ref_name="",
est_name="",
show_plot=False,
save_plot=None,
save_results=None,
no_warnings=False,
serialize_plot=None):
from evo.algorithms import metrics
from evo.algorithms import filters
from evo.algorithms import trajectory
from evo.tools import file_interface
from evo.tools.settings import SETTINGS
if not bins or not rel_tols:
raise RuntimeError(
"bins and tolerances must have more than one element")
if len(bins) != len(rel_tols):
raise RuntimeError(
"bins and tolerances must have the same number of elements")
if mode in {"speed", "angular_speed"
} and traj_est is trajectory.PosePath3D:
raise RuntimeError("timestamps are required for mode: " + mode)
bin_unit = None
if mode == "speed":
bin_unit = metrics.VelUnit.meters_per_sec
elif mode == "path":
bin_unit = metrics.Unit.meters
elif mode == "angle":
bin_unit = metrics.Unit.degrees
elif mode == "angular_speed":
bin_unit = metrics.VelUnit.degrees_per_sec
rpe_unit = None
if pose_relation is metrics.PoseRelation.translation_part:
rpe_unit = metrics.Unit.meters
elif pose_relation is metrics.PoseRelation.rotation_angle_deg:
rpe_unit = metrics.Unit.degrees
elif pose_relation is metrics.PoseRelation.rotation_angle_rad:
rpe_unit = metrics.Unit.radians
correct_only_scale = correct_scale and not align
if align or correct_scale:
logging.debug(SEP)
if correct_only_scale:
logging.debug("correcting scale...")
else:
logging.debug("aligning using Umeyama's method..." + (
" (with scale correction)" if correct_scale else ""))
traj_est = trajectory.align_trajectory(traj_est, traj_ref,
correct_scale,
correct_only_scale)
results = []
for bin, rel_tol, in zip(bins, rel_tols):
logging.debug(SEP)
logging.info("calculating RPE for each sub-sequence of " + str(bin) +
" (" + bin_unit.value + ")")
tol = bin * rel_tol
id_pairs = []
if mode == "path":
id_pairs = filters.filter_pairs_by_path(traj_ref.poses_se3,
bin,
tol,
all_pairs=True)
elif mode == "angle":
id_pairs = filters.filter_pairs_by_angle(traj_ref.poses_se3,
bin,
tol,
degrees=True)
elif mode == "speed":
id_pairs = filters.filter_pairs_by_speed(traj_ref.poses_se3,
traj_ref.timestamps, bin,
tol)
elif mode == "angular_speed":
id_pairs = filters.filter_pairs_by_angular_speed(
traj_ref.poses_se3, traj_ref.timestamps, bin, tol, True)
if len(id_pairs) == 0:
raise RuntimeError("bin " + str(bin) + " (" + str(bin_unit.value) +
") " +
"produced empty index list - try other values")
# calculate RPE with all IDs (delta 1 frames)
data = (traj_ref, traj_est)
# the delta here has nothing to do with the bin - 1f delta just to use all poses of the bin
rpe_metric = metrics.RPE(pose_relation,
delta=1,
delta_unit=metrics.Unit.frames,
all_pairs=True)
rpe_metric.process_data(data, id_pairs)
mean = rpe_metric.get_statistic(metrics.StatisticsType.mean)
results.append(mean)
if SETTINGS.plot_usetex:
mode.replace("_", "\_")
title = "mean RPE w.r.t. " + pose_relation.value + "\nfor different " + mode + " sub-sequences"
if align and not correct_scale:
title += "\n(with SE(3) Umeyama alignment)"
elif align and correct_scale:
title += "\n(with Sim(3) Umeyama alignment)"
elif correct_only_scale:
title += "\n(scale corrected)"
else:
title += "\n(not aligned)"
logging.debug(SEP)
logging.info("\n" + title + "\n")
res_str = ""
for bin, result in zip(bins, results):
res_str += "{:>10}".format(str(bin) + "(" + bin_unit.value + ")")
res_str += "\t" + "{0:.6f}".format(result) + "\n"
logging.info(res_str)
if show_plot or save_plot or serialize_plot:
from evo.tools import plot
import matplotlib.pyplot as plt
plot_collection = plot.PlotCollection(title)
fig = plt.figure(figsize=(SETTINGS.plot_figsize[0],
SETTINGS.plot_figsize[1]))
plot.error_array(fig,
results,
x_array=bins,
name="mean RPE" +
(" (" + rpe_unit.value + ")") if rpe_unit else "",
marker="o",
title=title,
xlabel=mode + " sub-sequences " + " (" +
bin_unit.value + ")")
# info text
if SETTINGS.plot_info_text and est_name and ref_name:
ax = fig.gca()
ax.text(0,
-0.12,
"estimate: " + est_name + "\nreference: " + ref_name,
transform=ax.transAxes,
fontsize=8,
color="gray")
plt.title(title)
plot_collection.add_figure("raw", fig)
if show_plot:
plot_collection.show()
if save_plot:
plot_collection.export(save_plot,
confirm_overwrite=not no_warnings)
if serialize_plot:
logging.debug(SEP)
plot_collection.serialize(serialize_plot,
confirm_overwrite=not no_warnings)
rpe_statistics = {bin: result for bin, result in zip(bins, results)}
if save_results:
logging.debug(SEP)
# utility class to trick save_res_file
class Metric:
unit = rpe_unit
error = results
file_interface.save_res_file(save_results,
Metric,
rpe_statistics,
title,
ref_name,
est_name,
bins,
traj_ref,
traj_est,
xlabel=mode + " sub-sequences " + " (" +
bin_unit.value + ")",
confirm_overwrite=not no_warnings)
return rpe_statistics, results
def main_rpe(traj_ref,
traj_est,
pose_relation,
delta,
delta_unit,
rel_delta_tol=0.1,
all_pairs=False,
align=False,
correct_scale=False,
ref_name="",
est_name="",
show_plot=False,
save_plot=None,
plot_mode=None,
save_results=None,
no_warnings=False,
support_loop=False,
serialize_plot=None):
from evo.core import metrics, result
from evo.core import trajectory
from evo.tools import file_interface
from evo.tools.settings import SETTINGS
if (show_plot or save_plot or serialize_plot) and all_pairs:
raise metrics.MetricsException(
"all_pairs mode cannot be used with plotting functions")
only_scale = correct_scale and not align
if align or correct_scale:
logging.debug(SEP)
if only_scale:
logging.debug("correcting scale...")
else:
logging.debug("aligning using Umeyama's method..." + (
" (with scale correction)" if correct_scale else ""))
traj_est = trajectory.align_trajectory(traj_est, traj_ref,
correct_scale, only_scale)
logging.debug(SEP)
# calculate RPE
data = (traj_ref, traj_est)
rpe_metric = metrics.RPE(pose_relation, delta, delta_unit, rel_delta_tol,
all_pairs)
rpe_metric.process_data(data)
rpe_statistics = rpe_metric.get_all_statistics()
title = str(rpe_metric)
if align and not correct_scale:
title += "\n(with SE(3) Umeyama alignment)"
elif align and correct_scale:
title += "\n(with Sim(3) Umeyama alignment)"
elif only_scale:
title += "\n(scale corrected)"
else:
title += "\n(not aligned)"
rpe_result = result.from_metric(rpe_metric, title, ref_name, est_name)
logging.debug(SEP)
logging.info(rpe_result.pretty_str())
# restrict trajectories to delta ids
if support_loop:
# avoid overwriting if called repeatedly e.g. in Jupyter notebook
import copy
traj_ref = copy.deepcopy(traj_ref)
traj_est = copy.deepcopy(traj_est)
traj_ref.reduce_to_ids(rpe_metric.delta_ids)
traj_est.reduce_to_ids(rpe_metric.delta_ids)
if isinstance(traj_est, trajectory.PoseTrajectory3D) and not all_pairs:
seconds_from_start = [
t - traj_est.timestamps[0] for t in traj_est.timestamps
]
rpe_result.add_np_array("seconds_from_start", seconds_from_start)
else:
seconds_from_start = None
if show_plot or save_plot or serialize_plot and not all_pairs:
from evo.tools import plot
import matplotlib.pyplot as plt
logging.debug(SEP)
logging.debug("plotting results... ")
fig1 = plt.figure(figsize=(SETTINGS.plot_figsize[0],
SETTINGS.plot_figsize[1]))
# metric values
plot.error_array(
fig1,
rpe_metric.error,
x_array=seconds_from_start,
statistics=rpe_statistics,
name="RPE" +
(" (" + rpe_metric.unit.value + ")") if rpe_metric.unit else "",
title=title,
xlabel="$t$ (s)" if seconds_from_start else "index")
# info text
if SETTINGS.plot_info_text and est_name and ref_name:
ax = fig1.gca()
ax.text(0,
-0.12,
"estimate: " + est_name + "\nreference: " + ref_name,
transform=ax.transAxes,
fontsize=8,
color="gray")
# trajectory colormapped
fig2 = plt.figure(figsize=(SETTINGS.plot_figsize[0],
SETTINGS.plot_figsize[1]))
plot_mode = plot_mode if plot_mode is not None else plot.PlotMode.xyz
ax = plot.prepare_axis(fig2, plot_mode)
plot.traj(ax,
plot_mode,
traj_ref,
'--',
'gray',
'reference',
alpha=0 if SETTINGS.plot_hideref else 1)
plot.traj_colormap(ax,
traj_est,
rpe_metric.error,
plot_mode,
min_map=rpe_statistics["min"],
max_map=rpe_statistics["max"],
title="RPE mapped onto trajectory")
fig2.axes.append(ax)
plot_collection = plot.PlotCollection(title)
plot_collection.add_figure("raw", fig1)
plot_collection.add_figure("map", fig2)
if show_plot:
plot_collection.show()
if save_plot:
plot_collection.export(save_plot,
confirm_overwrite=not no_warnings)
if serialize_plot:
logging.debug(SEP)
plot_collection.serialize(serialize_plot,
confirm_overwrite=not no_warnings)
if save_results:
logging.debug(SEP)
if SETTINGS.save_traj_in_zip:
rpe_result.add_trajectory("traj_ref", traj_ref)
rpe_result.add_trajectory("traj_est", traj_est)
file_interface.save_res_file(save_results,
rpe_result,
confirm_overwrite=not no_warnings)
return rpe_result
def plot_multi(args, result, traj_ref_list, traj_est_list):
from evo.tools import plot
from evo.tools.settings import SETTINGS
import matplotlib.pyplot as plt
import numpy as np
logger.debug(SEP)
logger.debug("Plotting results... ")
plot_mode = plot.PlotMode(args.plot_mode)
figs = []
# Plot the raw metric values.
error_array_comb = np.array([])
for i in range(len(result)):
figs.append(plt.figure(figsize=SETTINGS.plot_figsize))
if "seconds_from_start" in result[i].np_arrays:
seconds_from_start = result[i].np_arrays["seconds_from_start"]
else:
seconds_from_start = None
plot.error_array(
figs[i], result[i].np_arrays["error_array"], x_array=seconds_from_start,
statistics={
s: result[i].stats[s]
for s in SETTINGS.plot_statistics if s not in ("min", "max")
}, name=result[i].info["label"], title=result[i].info["title"],
xlabel="$t$ (s)" if seconds_from_start else "index")
# Plot the values color-mapped onto the trajectory.
figs.append(plt.figure(figsize=SETTINGS.plot_figsize))
ax = plot.prepare_axis(figs[-1], plot_mode)
if args.ros_map_yaml:
plot.ros_map(ax, args.ros_map_yaml, plot_mode)
if args.plot_colormap_min is None:
args.plot_colormap_min = min([result[i].stats["min"] for i in range(len(result))])
if args.plot_colormap_max is None:
args.plot_colormap_max = max([result[i].stats["max"] for i in range(len(result))])
if args.plot_colormap_max_percentile is not None:
args.plot_colormap_max = np.percentile(
error_array_comb, args.plot_colormap_max_percentile)
traj_est_list_comb = np.array([])
for i in range(len(result)):
plot.traj(ax, plot_mode, traj_ref_list[i], style=SETTINGS.plot_reference_linestyle,
color=multirobot_reference_color[i], label='reference'+str(i),
alpha=0.8)
plot.draw_coordinate_axes(ax, traj_ref_list[i], plot_mode,
SETTINGS.plot_axis_marker_scale)
plot.draw_coordinate_axes(ax, traj_est_list[i], plot_mode,
SETTINGS.plot_axis_marker_scale)
figs[-1].axes.append(ax)
plot.traj_colormap_multi(ax, traj_est_list, [r.np_arrays["error_array"] for r in result],
plot_mode, min_map=args.plot_colormap_min,
max_map=args.plot_colormap_max,
title="Error mapped onto trajectory")
plot_collection = plot.PlotCollection("Multi-robot APE analysis")
for i in range(len(result)):
plot_collection.add_figure("raw" + str(i), figs[i])
plot_collection.add_figure("map", figs[-1])
if args.plot:
plot_collection.show()
if args.save_plot:
plot_collection.export(args.save_plot,
confirm_overwrite=not args.no_warnings)
if args.serialize_plot:
logger.debug(SEP)
plot_collection.serialize(args.serialize_plot,
confirm_overwrite=not args.no_warnings)
def run_analysis(traj_ref_path,
traj_est_path,
segments,
save_results,
display_plot,
save_plots,
save_folder,
confirm_overwrite=False,
dataset_name="",
discard_n_start_poses=0,
discard_n_end_poses=0):
""" Run analysis on given trajectories, saves plots on given path:
:param traj_ref_path: path to the reference (ground truth) trajectory.
:param traj_est_path: path to the estimated trajectory.
:param save_results: saves APE, and RPE per segment results.
:param save_plots: whether to save the plots.
:param save_folder: where to save the plots.
:param confirm_overwrite: whether to confirm overwriting plots or not.
:param dataset_name: optional param, to allow setting the same scale on different plots.
"""
# Load trajectories.
from evo.tools import file_interface
traj_ref = None
try:
traj_ref = file_interface.read_euroc_csv_trajectory(
traj_ref_path) # TODO make it non-euroc specific.
except file_interface.FileInterfaceException as e:
raise Exception(
"\033[91mMissing ground truth csv! \033[93m {}.".format(e))
traj_est = None
try:
traj_est = file_interface.read_swe_csv_trajectory(traj_est_path)
except file_interface.FileInterfaceException as e:
log.info(e)
raise Exception("\033[91mMissing vio output csv.\033[99m")
evt.print_purple("Registering trajectories")
traj_ref, traj_est = sync.associate_trajectories(traj_ref, traj_est)
evt.print_purple("Aligning trajectories")
traj_est = trajectory.align_trajectory(
traj_est,
traj_ref,
correct_scale=False,
discard_n_start_poses=int(discard_n_start_poses),
discard_n_end_poses=int(discard_n_end_poses))
num_of_poses = traj_est.num_poses
traj_est.reduce_to_ids(
range(int(discard_n_start_poses),
int(num_of_poses - discard_n_end_poses), 1))
traj_ref.reduce_to_ids(
range(int(discard_n_start_poses),
int(num_of_poses - discard_n_end_poses), 1))
results = dict()
evt.print_purple("Calculating APE translation part")
data = (traj_ref, traj_est)
ape_metric = metrics.APE(metrics.PoseRelation.translation_part)
ape_metric.process_data(data)
ape_result = ape_metric.get_result()
results["absolute_errors"] = ape_result
log.info(ape_result.pretty_str(info=True))
# TODO(Toni): Save RPE computation results rather than the statistics
# you can compute statistics later...
evt.print_purple("Calculating RPE translation part for plotting")
rpe_metric_trans = metrics.RPE(metrics.PoseRelation.translation_part, 1.0,
metrics.Unit.frames, 0.0, False)
rpe_metric_trans.process_data(data)
rpe_stats_trans = rpe_metric_trans.get_all_statistics()
log.info("mean: %f" % rpe_stats_trans["mean"])
evt.print_purple("Calculating RPE rotation angle for plotting")
rpe_metric_rot = metrics.RPE(metrics.PoseRelation.rotation_angle_deg, 1.0,
metrics.Unit.frames, 1.0, False)
rpe_metric_rot.process_data(data)
rpe_stats_rot = rpe_metric_rot.get_all_statistics()
log.info("mean: %f" % rpe_stats_rot["mean"])
results["relative_errors"] = dict()
# Read segments file
for segment in segments:
results["relative_errors"][segment] = dict()
evt.print_purple("RPE analysis of segment: %d" % segment)
evt.print_lightpurple("Calculating RPE segment translation part")
rpe_segment_metric_trans = metrics.RPE(
metrics.PoseRelation.translation_part, float(segment),
metrics.Unit.meters, 0.01, True)
rpe_segment_metric_trans.process_data(data)
rpe_segment_stats_trans = rpe_segment_metric_trans.get_all_statistics()
results["relative_errors"][segment][
"rpe_trans"] = rpe_segment_stats_trans
# print(rpe_segment_stats_trans)
# print("mean:", rpe_segment_stats_trans["mean"])
evt.print_lightpurple("Calculating RPE segment rotation angle")
rpe_segment_metric_rot = metrics.RPE(
metrics.PoseRelation.rotation_angle_deg, float(segment),
metrics.Unit.meters, 0.01, True)
rpe_segment_metric_rot.process_data(data)
rpe_segment_stats_rot = rpe_segment_metric_rot.get_all_statistics()
results["relative_errors"][segment]["rpe_rot"] = rpe_segment_stats_rot
# print(rpe_segment_stats_rot)
# print("mean:", rpe_segment_stats_rot["mean"])
if save_results:
# Save results file
results_file = os.path.join(save_folder, 'results.yaml')
evt.print_green("Saving analysis results to: %s" % results_file)
with open(results_file, 'w') as outfile:
if confirm_overwrite:
if evt.user.check_and_confirm_overwrite(results_file):
outfile.write(yaml.dump(results, default_flow_style=False))
else:
log.info("Not overwritting results.")
else:
outfile.write(yaml.dump(results, default_flow_style=False))
# For each segment in segments file
# Calculate rpe with delta = segment in meters with all-pairs set to True
# Calculate max, min, rmse, mean, median etc
# Plot boxplot, or those cumulative figures you see in evo (like demographic plots)
if display_plot or save_plots:
evt.print_green("Plotting:")
log.info(dataset_name)
plot_collection = plot.PlotCollection("Example")
# metric values
fig_1 = plt.figure(figsize=(8, 8))
ymax = -1
if dataset_name is not "" and FIX_MAX_Y:
ymax = Y_MAX_APE_TRANS[dataset_name]
ape_statistics = ape_metric.get_all_statistics()
plot.error_array(
fig_1,
ape_metric.error,
statistics=ape_statistics,
name="APE translation",
title="" #str(ape_metric)
,
xlabel="Keyframe index [-]",
ylabel="APE translation [m]",
y_min=0.0,
y_max=ymax)
plot_collection.add_figure("APE_translation", fig_1)
# trajectory colormapped with error
fig_2 = plt.figure(figsize=(8, 8))
plot_mode = plot.PlotMode.xy
ax = plot.prepare_axis(fig_2, plot_mode)
plot.traj(ax, plot_mode, traj_ref, '--', 'gray', 'reference')
plot.traj_colormap(ax,
traj_est,
ape_metric.error,
plot_mode,
min_map=0.0,
max_map=math.ceil(ape_statistics['max'] * 10) / 10,
title="ATE mapped onto trajectory [m]")
plot_collection.add_figure("APE_translation_trajectory_error", fig_2)
# RPE
## Trans
### metric values
fig_3 = plt.figure(figsize=(8, 8))
if dataset_name is not "" and FIX_MAX_Y:
ymax = Y_MAX_RPE_TRANS[dataset_name]
plot.error_array(
fig_3,
rpe_metric_trans.error,
statistics=rpe_stats_trans,
name="RPE translation",
title="" #str(rpe_metric_trans)
,
xlabel="Keyframe index [-]",
ylabel="RPE translation [m]",
y_max=ymax)
plot_collection.add_figure("RPE_translation", fig_3)
### trajectory colormapped with error
fig_4 = plt.figure(figsize=(8, 8))
plot_mode = plot.PlotMode.xy
ax = plot.prepare_axis(fig_4, plot_mode)
traj_ref_trans = copy.deepcopy(traj_ref)
traj_ref_trans.reduce_to_ids(rpe_metric_trans.delta_ids)
traj_est_trans = copy.deepcopy(traj_est)
traj_est_trans.reduce_to_ids(rpe_metric_trans.delta_ids)
plot.traj(ax, plot_mode, traj_ref_trans, '--', 'gray', 'Reference')
plot.traj_colormap(
ax,
traj_est_trans,
rpe_metric_trans.error,
plot_mode,
min_map=0.0,
max_map=math.ceil(rpe_stats_trans['max'] * 10) / 10,
title="RPE translation error mapped onto trajectory [m]")
plot_collection.add_figure("RPE_translation_trajectory_error", fig_4)
## Rot
### metric values
fig_5 = plt.figure(figsize=(8, 8))
if dataset_name is not "" and FIX_MAX_Y:
ymax = Y_MAX_RPE_ROT[dataset_name]
plot.error_array(
fig_5,
rpe_metric_rot.error,
statistics=rpe_stats_rot,
name="RPE rotation error",
title="" #str(rpe_metric_rot)
,
xlabel="Keyframe index [-]",
ylabel="RPE rotation [deg]",
y_max=ymax)
plot_collection.add_figure("RPE_rotation", fig_5)
### trajectory colormapped with error
fig_6 = plt.figure(figsize=(8, 8))
plot_mode = plot.PlotMode.xy
ax = plot.prepare_axis(fig_6, plot_mode)
traj_ref_rot = copy.deepcopy(traj_ref)
traj_ref_rot.reduce_to_ids(rpe_metric_rot.delta_ids)
traj_est_rot = copy.deepcopy(traj_est)
traj_est_rot.reduce_to_ids(rpe_metric_rot.delta_ids)
plot.traj(ax, plot_mode, traj_ref_rot, '--', 'gray', 'Reference')
plot.traj_colormap(
ax,
traj_est_rot,
rpe_metric_rot.error,
plot_mode,
min_map=0.0,
max_map=math.ceil(rpe_stats_rot['max'] * 10) / 10,
title="RPE rotation error mapped onto trajectory [deg]")
plot_collection.add_figure("RPE_rotation_trajectory_error", fig_6)
if display_plot:
evt.print_green("Displaying plots.")
plot_collection.show()
if save_plots:
evt.print_green("Saving plots to: ")
log.info(save_folder)
# Config output format (pdf, eps, ...) using evo_config...
plot_collection.export(os.path.join(save_folder, "plots.eps"),
False)
plot_collection.export(os.path.join(save_folder, "plots.pdf"),
False)
for pose_relation in metrics.PoseRelation:
print("\n------------------------------------------------------------------\n")
data = (traj_ref, traj_est)
start = time.clock()
ape_metric = metrics.APE(pose_relation)
ape_metric.process_data(data)
ape_stats = ape_metric.get_all_statistics()
stop = time.clock()
ape_time = stop-start
print("APE statistics w.r.t. " + ape_metric.pose_relation.value + ": ")
pretty_printer.pprint(ape_stats)
print("\nelapsed time for running the APE algorithm (seconds):\t", "{0:.6f}".format(ape_time))
print("elapsed time for trajectory loading and APE (seconds):\t", "{0:.6f}".format(load_time+ape_time))
if show_plots:
plot.error_array(ape_metric.error, statistics=ape_stats, name="APE w.r.t. " + ape_metric.pose_relation.value)
if show_plots:
plt.show()
print("\n------------------------------------------------------------------")
print("------------------------------------------------------------------\n")
print("calling offical TUM ATE script for comparison...")
cmd = ["tum_benchmark_tools/evaluate_ate.py", ref_file, est_file, "--max_difference", str(max_diff), "--verbose"]
time_cmd = timeit(stmt="sp.call("+str(cmd)+")", setup="import subprocess as sp", number=1)
print("\nelapsed time for full TUM ATE script (seconds):\t", "{0:.6f}".format(time_cmd))
def plot(args, result, traj_ref, traj_est):
from evo.tools import plot
from evo.tools.settings import SETTINGS
import matplotlib.pyplot as plt
import numpy as np
logger.debug(SEP)
logger.debug("Plotting results... ")
plot_mode = plot.PlotMode(args.plot_mode)
# Plot the raw metric values.
fig1 = plt.figure(figsize=SETTINGS.plot_figsize)
if "seconds_from_start" in result.np_arrays:
seconds_from_start = result.np_arrays["seconds_from_start"]
else:
seconds_from_start = None
plot.error_array(
fig1, result.np_arrays["error_array"],
x_array=seconds_from_start,
statistics=result.stats,
name=result.info["label"],
title=result.info["title"],
xlabel="$t$ (s)" if seconds_from_start else "index")
# Plot the values color-mapped onto the trajectory.
fig2 = plt.figure(figsize=SETTINGS.plot_figsize)
ax = plot.prepare_axis(fig2, plot_mode)
plot.traj(ax, plot_mode, traj_ref, '--', 'black', 'reference',
alpha=0.0 if SETTINGS.plot_hideref else 0.5)
if args.plot_colormap_min is None:
args.plot_colormap_min = result.stats["min"]
if args.plot_colormap_max is None:
args.plot_colormap_max = result.stats["max"]
if args.plot_colormap_max_percentile is not None:
args.plot_colormap_max = np.percentile(
result.np_arrays["error_array"],
args.plot_colormap_max_percentile)
plot.traj_colormap(
ax, traj_est, result.np_arrays["error_array"],
plot_mode,
min_map=args.plot_colormap_min,
max_map=args.plot_colormap_max,
title="Error mapped onto trajectory")
fig2.axes.append(ax)
plot_collection = plot.PlotCollection(result.info["title"])
plot_collection.add_figure("raw", fig1)
plot_collection.add_figure("map", fig2)
if args.plot:
plot_collection.show()
if args.save_plot:
plot_collection.export(
args.save_plot, confirm_overwrite=not args.no_warnings)
if args.serialize_plot:
logger.debug(SEP)
plot_collection.serialize(
args.serialize_plot, confirm_overwrite=not args.no_warnings)
