def path_following( controller, par, path, x, y, psi, velocity, Delta_l_r = 0.0, Delta_l_r_dot = None, psi_dot = None, velocity_dot = None, Delta_l_r_dotdot = None, Ts=0.01 ): """ Basic steering control for path tracking and user-defined lateral error compensation Implements steering control for exact path following. Assumed is a kinematic bicycle model. Herein, the steering angle (delta) is the control variable. The variables x, y, psi, and velocity are measurements taken from the controlled system. The lateral offset Delta_l_r to the path is the reference for control. The optional signal Delta_l_r_dot describes the time derivative of Delta_l_r. controller - callback function that defines the error compensating controller par - parameters that are passed to the callback Ts - the sampling time Return values ------------- results = {} results['x_r'] # the current x-position of the closest point on the reference path results['y_r'] # the current y-position of the closest point on the reference path results['v_star'] # the current velocity of the closest point on the reference path results['d_star'] # the current distance parameter of the closest point on the reference path results['psi_r'] # the current path-tangent orientation angle in the closest point on the reference path results['psi_r_dot'] # the time derivative of psi_r results['Delta_l_r'] # the reference to the distance to the path results['Delta_l_r_dot'] # optional: the time derivative of Delta_l_r results['Delta_l'] # the distance to the closest point on the reference path results['Delta_u'] # small steering delta results['delta'] # the requested steering angle / the control variable in case Delta_l_r_dot, psi_dot, velocity_dot, and Delta_l_r_dotdot are given the steering derivatives can be computed. results['Delta_u_dot'] # the derivative of Delta_u results['delta_dot'] # the derivative of delta_dot """ index_head, _ = path_horizon_head_index(path) # structure for output signals results = dy.structure() # track the evolution of the closest point on the path to the vehicles position minimal_number_of_path_samples_to_start = 5 # depends on tracker(); should be at least 2 with dy.sub_if( index_head > minimal_number_of_path_samples_to_start, subsystem_name='tracker' ) as system: tracking_results = tracker(path, x, y) system.set_outputs( tracking_results.to_list() ) tracking_results.replace_signals( system.outputs ) output_valid = tracking_results['minimal_distance_reached'] need_more_path_input_data = tracking_results['reached_the_end_of_currently_available_path_data'] # in case the lookup was successful, run further operations on the path # to generate references and run the controller. with dy.sub_if( output_valid, prevent_output_computation=False, subsystem_name='controller') as system: # ps - path sample ps = track_projection_on_path(path, x, y, tracking_results = tracking_results) # # project the vehicle velocity onto the path yielding v_star # # Used formula inside project_velocity_on_path: # v_star = d d_star / dt = v * cos( Delta_u ) / ( 1 - Delta_l * K(d_star) ) # Delta_u = dy.signal() # feedback from control v_star = project_velocity_on_path(velocity, Delta_u, ps['Delta_l'], ps['K_r']) # # compute an enhanced (less noise) signal for the path orientation psi_r by integrating the # curvature profile and fusing the result with ps['psi_r'] to mitigate the integration drift. # psi_r, psi_r_dot = compute_path_orientation_from_curvature( Ts, v_star, ps['psi_r'], ps['K_r'], L=1.0 ) # # controller callback # references = { 'Delta_l_r' : Delta_l_r, 'Delta_l_r_dot' : Delta_l_r_dot, 'Delta_l_r_dotdot' : Delta_l_r_dotdot } # Delta_l_dot might be further computed which improves the accuracy of the derivatives # in case of strong feedback control activity. Delta_l_dot = None # TODO: implement measurements = { 'velocity' : velocity, 'velocity_dot' : velocity_dot, 'psi' : psi, 'psi_dot' : psi_dot, 'Delta_l' : ps['Delta_l'], 'Delta_l_dot' : Delta_l_dot } u, u_dot = controller( references, measurements, par ) # # path tracking # # resulting lateral model u --> Delta_l : 1/s # Delta_u << dy.asin( dy.saturate(u / velocity, -0.99, 0.99) ) delta = dy.unwrap_angle(angle=psi_r - psi + Delta_u, normalize_around_zero = True) # compute the derivatives of the steering angle (steering rate) if psi_dot is not None and Delta_l_r_dot is not None and velocity_dot is not None and Delta_l_r_dotdot is not None: Delta_u_dot = dy.cos( u / velocity ) * ( velocity * u_dot - velocity_dot * u ) / ( velocity*velocity ) delta_dot = psi_r_dot - psi_dot + Delta_u_dot results['Delta_u_dot'] = Delta_u_dot results['delta_dot'] = delta_dot # collect resulting signals results['x_r'] = ps['x_r'] # the current x-position of the closest point on the reference path results['y_r'] = ps['y_r'] # the current y-position of the closest point on the reference path results['v_star'] = v_star # the current velocity of the closest point on the reference path results['d_star'] = ps['d_star'] # the current distance parameter of the closest point on the reference path results['psi_r'] = psi_r # the current path-tangent orientation angle in the closest point on the reference path results['K_r'] = ps['K_r'] # the curvature results['psi_r_dot'] = psi_r_dot # the time derivative of psi_r results['Delta_u'] = Delta_u # small steering delta results['delta'] = delta # the requested steering angle / the control variable results['Delta_l'] = ps['Delta_l'] # the distance to the closest point on the reference path results['Delta_l_dot'] = dy.float64(math.nan) # d/dt Delta_l TODO: implement # results['line_tracking_internals'] = ps['internals'] # return system.set_outputs( results.to_list() ) results.replace_signals( system.outputs ) results['tracked_index'] = tracking_results['tracked_index'] results['Delta_l_r'] = Delta_l_r # the reference to the distance to the path results['need_more_path_input_data'] = need_more_path_input_data results['output_valid'] = output_valid results['read_position'] = results['tracked_index'] + 1 results['minimal_read_position'] = results['read_position'] - 100 return results
# track the evolution of the closest point on the path to the vehicles position d_star, x_r, y_r, psi_r, K_r, Delta_l, tracked_index, Delta_index = track_projection_on_path( path, x, y) # reference for the lateral distance Delta_l_r = dy.float64(0.0) # zero in this example dy.append_output(Delta_l_r, 'Delta_l_r') # feedback control u = dy.PID_controller(r=Delta_l_r, y=Delta_l, Ts=0.01, kp=k_p) # path tracking # resulting lateral model u --> Delta_l : 1/s Delta_u = dy.asin(dy.saturate(u / velocity, -0.99, 0.99)) delta_star = psi_r - psi steering = delta_star + Delta_u steering = dy.unwrap_angle(angle=steering, normalize_around_zero=True) dy.append_output(Delta_u, 'Delta_u') dy.append_output(delta_star, 'delta_star') # # The model of the vehicle including a disturbance # # model the disturbance disturbance_transient = np.concatenate( (cosra(50, 0, 1.0), co(10, 1.0), cosra(50, 1.0, 0))) steering_disturbance, i = dy.play(
def path_following_controller_P( path, x, y, psi, velocity, Delta_l_r = 0.0, Delta_l_r_dot = None, k_p=2.0, Ts=0.01, psi_dot = None, velocity_dot = None, Delta_l_r_dotdot = None, Delta_l_dot = None ): """ Basic steering control for path tracking using proportional lateral error compensation Path following steering control for exact path following and P-control to control the lateral distance to the path are combined. Controlls a kinematic bicycle model (assumption) to follow the given path. Herein, the steering angle delta is the control variable. The variables x, y, psi, and velocity are measurements taken from the controlled system. The lateral offset Delta_l_r to the path is the reference. The optional signal Delta_l_r_dot describes the time derivative of Delta_l_r. Ts - the sampling time Return values ------------- results = {} results['x_r'] # the current x-position of the closest point on the reference path results['y_r'] # the current y-position of the closest point on the reference path results['v_star'] # the current velocity of the closest point on the reference path results['d_star'] # the current distance parameter of the closest point on the reference path results['psi_r'] # the current path-tangent orientation angle in the closest point on the reference path results['psi_r_dot'] # the time derivative of psi_r results['Delta_l_r'] # the reference to the distance to the path results['Delta_l_r_dot'] # optional: the time derivative of Delta_l_r results['Delta_l'] # the distance to the closest point on the reference path results['Delta_u'] # small steering delta results['delta'] # the requested steering angle / the control variable in case Delta_l_r_dot, psi_dot, velocity_dot, and Delta_l_r_dotdot are given the steering derivatives can be computed. results['Delta_u_dot'] # the derivative of Delta_u results['delta_dot'] # the derivative of delta_dot Optionally, Delta_l_dot might be further given which improves the accuracy of the derivatives in case of strong feedback control activity. """ index_head, _ = path_horizon_head_index(path) # structure for output signals results = dy.structure() # track the evolution of the closest point on the path to the vehicles position with dy.sub_if( index_head > 10 ) as system: tracking_results = tracker(path, x, y) system.set_outputs( tracking_results.to_list() ) tracking_results.replace_signals( system.outputs ) output_valid = tracking_results['minimal_distance_reached'] need_more_path_input_data = tracking_results['reached_the_end_of_currently_available_path_data'] # position_on_path_found = dy.boolean(True) with dy.sub_if( output_valid, prevent_output_computation=False, subsystem_name='controller') as system: # ps - path sample ps = track_projection_on_path(path, x, y, tracking_results = tracking_results) # # project the vehicle velocity onto the path yielding v_star # # Used formula inside project_velocity_on_path: # v_star = d d_star / dt = v * cos( Delta_u ) / ( 1 - Delta_l * K(d_star) ) # Delta_u = dy.signal() # feedback from control v_star = project_velocity_on_path(velocity, Delta_u, ps['Delta_l'], ps['K_r']) # # compute an enhanced (less noise) signal for the path orientation psi_r by integrating the # curvature profile and fusing the result with ps['psi_r'] to mitigate the integration drift. # psi_r, psi_r_dot = compute_path_orientation_from_curvature( Ts, v_star, ps['psi_r'], ps['K_r'], L=1.0 ) # feedback control u_fb = k_p * (Delta_l_r - ps['Delta_l']) if Delta_l_r_dot is not None: u = Delta_l_r_dot + u_fb else: u = u_fb # path tracking # resulting lateral model u --> Delta_l : 1/s Delta_u << dy.asin( dy.saturate(u / velocity, -0.99, 0.99) ) delta = dy.unwrap_angle(angle=psi_r - psi + Delta_u, normalize_around_zero = True) # compute the derivatives of the steering angle (steering rate) if psi_dot is not None and Delta_l_r_dot is not None and velocity_dot is not None and Delta_l_r_dotdot is not None: if Delta_l_dot is None: u_dot = Delta_l_r_dotdot # + 0 neglect numerical random walk error compensation else: u_dot = Delta_l_r_dotdot + Delta_l_dot Delta_u_dot = dy.cos( u / velocity ) * ( velocity * u_dot - velocity_dot * u ) / ( velocity*velocity ) delta_dot = psi_r_dot - psi_dot + Delta_u_dot results['Delta_u_dot'] = Delta_u_dot results['delta_dot'] = delta_dot # collect resulting signals results['x_r'] = ps['x_r'] # the current x-position of the closest point on the reference path results['y_r'] = ps['y_r'] # the current y-position of the closest point on the reference path results['v_star'] = v_star # the current velocity of the closest point on the reference path results['d_star'] = ps['d_star'] # the current distance parameter of the closest point on the reference path results['psi_r'] = psi_r # the current path-tangent orientation angle in the closest point on the reference path results['psi_r_dot'] = psi_r_dot # the time derivative of psi_r results['Delta_l'] = ps['Delta_l'] # the distance to the closest point on the reference path results['Delta_u'] = Delta_u # small steering delta results['delta'] = delta # the requested steering angle / the control variable # results['line_tracking_internals'] = ps['internals'] # return system.set_outputs( results.to_list() ) results.replace_signals( system.outputs ) results['tracked_index'] = tracking_results['tracked_index'] results['Delta_l_r'] = Delta_l_r # the reference to the distance to the path results['need_more_path_input_data'] = need_more_path_input_data results['output_valid'] = output_valid results['read_position'] = results['tracked_index'] + 1 results['minimal_read_position'] = results['read_position'] - 100 return results