Newer
Older
'''This class will log 1d array in Nd matrix from device and qualisys object'''
import numpy as np
from datetime import datetime as datetime
from time import time
from utils_mpc import quaternionToRPY
class LoggerControl():
Pierre-Alexandre Leziart
committed
def __init__(self, dt, N0_gait, joystick=None, estimator=None, loop=None, gait=None, statePlanner=None,
Pierre-Alexandre Leziart
committed
footstepPlanner=None, footTrajectoryGenerator=None, logSize=60e3, ringBuffer=False):
self.ringBuffer = ringBuffer
logSize = np.int(logSize)
self.logSize = logSize
self.i = 0
self.dt = dt
# Allocate the data:
# Joystick
self.joy_v_ref = np.zeros([logSize, 6]) # reference velocity of the joystick
# Estimator
self.esti_feet_status = np.zeros([logSize, 4]) # input feet status (contact or not)
self.esti_feet_goals = np.zeros([logSize, 3, 4]) # input feet goals (desired on the ground)
self.esti_q_filt = np.zeros([logSize, 19]) # output position
self.esti_v_filt = np.zeros([logSize, 18]) # output velocity
self.esti_v_secu = np.zeros([logSize, 12]) # filtered output velocity for security check
self.esti_FK_lin_vel = np.zeros([logSize, 3]) # estimated velocity of the base with FK
self.esti_FK_xyz = np.zeros([logSize, 3]) # estimated position of the base with FK
self.esti_xyz_mean_feet = np.zeros([logSize, 3]) # average of feet goals
self.esti_HP_x = np.zeros([logSize, 3]) # x input of the velocity complementary filter
self.esti_HP_dx = np.zeros([logSize, 3]) # dx input of the velocity complementary filter
self.esti_HP_alpha = np.zeros([logSize, 3]) # alpha parameter of the velocity complementary filter
self.esti_HP_filt_x = np.zeros([logSize, 3]) # filtered output of the velocity complementary filter
self.esti_LP_x = np.zeros([logSize, 3]) # x input of the position complementary filter
self.esti_LP_dx = np.zeros([logSize, 3]) # dx input of the position complementary filter
self.esti_LP_alpha = np.zeros([logSize, 3]) # alpha parameter of the position complementary filter
self.esti_LP_filt_x = np.zeros([logSize, 3]) # filtered output of the position complementary filter
Pierre-Alexandre Leziart
committed
self.esti_kf_X = np.zeros([logSize, 18]) # state of the Kalman filter
self.esti_kf_Z = np.zeros([logSize, 16]) # measurement for the Kalman filter
# Loop
self.loop_o_q_int = np.zeros([logSize, 19]) # position in world frame (esti_q_filt + dt * loop_o_v)
self.loop_o_v = np.zeros([logSize, 18]) # estimated velocity in world frame
Pierre-Alexandre Leziart
committed
# Gait
Pierre-Alexandre Leziart
committed
self.planner_gait = np.zeros([logSize, N0_gait, 4]) # Gait sequence
Pierre-Alexandre Leziart
committed
self.planner_is_static = np.zeros([logSize]) # if the planner is in static mode or not
self.planner_q_static = np.zeros([logSize, 19]) # position in static mode (4 stance phase)
self.planner_RPY_static = np.zeros([logSize, 3]) # RPY orientation in static mode (4 stance phase)
Pierre-Alexandre Leziart
committed
# State planner
Pierre-Alexandre Leziart
committed
if statePlanner is not None:
self.planner_xref = np.zeros([logSize, 12, 1+statePlanner.getNSteps()]) # Reference trajectory
Pierre-Alexandre Leziart
committed
# Footstep planner
Pierre-Alexandre Leziart
committed
if gait is not None:
self.planner_fsteps = np.zeros([logSize, gait.getCurrentGait().shape[0], 12]) # Reference footsteps position
Pierre-Alexandre Leziart
committed
self.planner_h_ref = np.zeros([logSize]) # reference height of the planner
# Foot Trajectory Generator
self.planner_goals = np.zeros([logSize, 3, 4]) # 3D target feet positions
self.planner_vgoals = np.zeros([logSize, 3, 4]) # 3D target feet velocities
self.planner_agoals = np.zeros([logSize, 3, 4]) # 3D target feet accelerations
# Model Predictive Control
# output vector of the MPC (next state + reference contact force)
Pierre-Alexandre Leziart
committed
if statePlanner is not None:
self.mpc_x_f = np.zeros([logSize, 24, statePlanner.getNSteps()])
# Whole body control
self.wbc_x_f = np.zeros([logSize, 24]) # input vector of the WBC (next state + reference contact force)
self.wbc_P = np.zeros([logSize, 12]) # proportionnal gains of the PD+
self.wbc_D = np.zeros([logSize, 12]) # derivative gains of the PD+
self.wbc_q_des = np.zeros([logSize, 12]) # desired position of actuators
self.wbc_v_des = np.zeros([logSize, 12]) # desired velocity of actuators
self.wbc_tau_ff = np.zeros([logSize, 12]) # feedforward torques computed by the WBC
self.wbc_f_ctc = np.zeros([logSize, 12]) # contact forces computed by the WBC
self.wbc_feet_pos = np.zeros([logSize, 3, 4]) # current feet positions according to WBC
self.wbc_feet_err = np.zeros([logSize, 3, 4]) # error between feet positions and their reference
self.wbc_feet_vel = np.zeros([logSize, 3, 4]) # current feet velocities according to WBC
self.wbc_feet_pos_invkin = np.zeros([logSize, 3, 4]) # current feet positions according to InvKin
self.wbc_feet_vel_invkin = np.zeros([logSize, 3, 4]) # current feet velocities according to InvKin
# Timestamps
self.tstamps = np.zeros(logSize)
Pierre-Alexandre Leziart
committed
def sample(self, joystick, estimator, loop, gait, statePlanner, footstepPlanner, footTrajectoryGenerator, wbc):
if (self.i >= self.logSize):
if self.ringBuffer:
self.i = 0
else:
return
# Logging from joystick
self.joy_v_ref[self.i] = joystick.v_ref[:, 0]
# Logging from estimator
self.esti_feet_status[self.i] = estimator.feet_status[:]
self.esti_feet_goals[self.i] = estimator.feet_goals
self.esti_q_filt[self.i] = estimator.q_filt[:, 0]
self.esti_v_filt[self.i] = estimator.v_filt[:, 0]
self.esti_v_secu[self.i] = estimator.v_secu[:]
self.esti_FK_lin_vel[self.i] = estimator.FK_lin_vel[:]
self.esti_FK_xyz[self.i] = estimator.FK_xyz[:]
self.esti_xyz_mean_feet[self.i] = estimator.xyz_mean_feet[:]
Pierre-Alexandre Leziart
committed
if not estimator.kf_enabled:
self.esti_HP_x[self.i] = estimator.filter_xyz_vel.x
self.esti_HP_dx[self.i] = estimator.filter_xyz_vel.dx
self.esti_HP_alpha[self.i] = estimator.filter_xyz_vel.alpha
self.esti_HP_filt_x[self.i] = estimator.filter_xyz_vel.filt_x
self.esti_LP_x[self.i] = estimator.filter_xyz_pos.x
self.esti_LP_dx[self.i] = estimator.filter_xyz_pos.dx
self.esti_LP_alpha[self.i] = estimator.filter_xyz_pos.alpha
self.esti_LP_filt_x[self.i] = estimator.filter_xyz_pos.filt_x
else:
self.esti_kf_X[self.i] = estimator.kf.X[:, 0]
self.esti_kf_Z[self.i] = estimator.Z[:, 0]
# Logging from the main loop
Pierre-Alexandre Leziart
committed
self.loop_o_q_int[self.i] = loop.q[:, 0]
self.loop_o_v[self.i] = loop.v[:, 0]
# Logging from the planner
Pierre-Alexandre Leziart
committed
# self.planner_q_static[self.i] = planner.q_static[:]
# self.planner_RPY_static[self.i] = planner.RPY_static[:, 0]
self.planner_xref[self.i] = statePlanner.getReferenceStates()
Pierre-Alexandre Leziart
committed
self.planner_fsteps[self.i] = footstepPlanner.getFootsteps()
self.planner_gait[self.i] = gait.getCurrentGait()
self.planner_goals[self.i] = footTrajectoryGenerator.getFootPosition()
self.planner_vgoals[self.i] = footTrajectoryGenerator.getFootVelocity()
self.planner_agoals[self.i] = footTrajectoryGenerator.getFootAcceleration()
self.planner_is_static[self.i] = gait.getIsStatic()
self.planner_h_ref[self.i] = loop.h_ref
# Logging from model predictive control
self.mpc_x_f[self.i] = loop.x_f_mpc
# Logging from whole body control
self.wbc_x_f[self.i] = loop.x_f_wbc
self.wbc_P[self.i] = loop.result.P
self.wbc_D[self.i] = loop.result.D
self.wbc_q_des[self.i] = loop.result.q_des
self.wbc_v_des[self.i] = loop.result.v_des
self.wbc_tau_ff[self.i] = loop.result.tau_ff
self.wbc_f_ctc[self.i] = wbc.f_with_delta[:, 0]
self.wbc_feet_pos[self.i] = wbc.feet_pos
self.wbc_feet_err[self.i] = wbc.feet_err
self.wbc_feet_vel[self.i] = wbc.feet_vel
self.wbc_feet_pos_invkin[self.i] = wbc.invKin.cpp_posf.transpose()
self.wbc_feet_vel_invkin[self.i] = wbc.invKin.cpp_vf.transpose()
# Logging timestamp
self.tstamps[self.i] = time()
self.i += 1
def processMocap(self, N, loggerSensors):
self.mocap_b_v = np.zeros([N, 3])
self.mocap_b_w = np.zeros([N, 3])
self.mocap_RPY = np.zeros([N, 3])
for i in range(N):
oRb = loggerSensors.mocapOrientationMat9[i]
"""from IPython import embed
embed()"""
self.mocap_b_v[i] = (oRb.transpose() @ loggerSensors.mocapVelocity[i].reshape((3, 1))).ravel()
self.mocap_b_w[i] = (oRb.transpose() @ loggerSensors.mocapAngularVelocity[i].reshape((3, 1))).ravel()
self.mocap_RPY[i] = quaternionToRPY(loggerSensors.mocapOrientationQuat[i])[:, 0]
def plotAll(self, loggerSensors):
from matplotlib import pyplot as plt
N = self.tstamps.shape[0]
t_range = np.array([k*self.dt for k in range(N)])
self.processMocap(N, loggerSensors)
index6 = [1, 3, 5, 2, 4, 6]
index12 = [1, 5, 9, 2, 6, 10, 3, 7, 11, 4, 8, 12]
"""plt.figure()
for i in range(4):
if i == 0:
ax0 = plt.subplot(2, 2, i+1)
else:
plt.subplot(2, 2, i+1, sharex=ax0)
switch = np.diff(self.esti_feet_status[:, i])
tmp = self.wbc_feet_pos[:-1, 2, i]
tmp_y = tmp[switch > 0]
tmp_x = t_range[:-1]
tmp_x = tmp_x[switch > 0]
plt.plot(tmp_x, tmp_y, linewidth=3)"""
lgd_X = ["FL", "FR", "HL", "HR"]
lgd_Y = ["Pos X", "Pos Y", "Pos Z"]
plt.figure()
for i in range(12):
if i == 0:
ax0 = plt.subplot(3, 4, index12[i])
else:
plt.subplot(3, 4, index12[i], sharex=ax0)
plt.plot(t_range, self.wbc_feet_pos[:, i % 3, np.int(i/3)], color='b', linewidth=3, marker='')
plt.plot(t_range, self.wbc_feet_err[:, i % 3, np.int(i/3)], color='g', linewidth=3, marker='')
plt.plot(t_range, self.planner_goals[:, i % 3, np.int(i/3)], color='r', linewidth=3, marker='')
plt.plot(t_range, self.wbc_feet_pos_invkin[:, i % 3, np.int(i/3)],
color='darkviolet', linewidth=3, linestyle="--", marker='')
Pierre-Alexandre Leziart
committed
plt.plot(t_range, self.planner_gait[:, 0, np.int(
i/3)] * np.max(self.wbc_feet_pos[:, i % 3, np.int(i/3)]), color='k', linewidth=3, marker='')
plt.legend([lgd_Y[i % 3] + " " + lgd_X[np.int(i/3)]+"", "error",
lgd_Y[i % 3] + " " + lgd_X[np.int(i/3)]+" Ref", "Contact state"], prop={'size': 8})
plt.suptitle("Measured & Reference feet positions (world frame)")
lgd_X = ["FL", "FR", "HL", "HR"]
lgd_Y = ["Vel X", "Vel Y", "Vel Z"]
plt.figure()
for i in range(12):
if i == 0:
ax0 = plt.subplot(3, 4, index12[i])
else:
plt.subplot(3, 4, index12[i], sharex=ax0)
plt.plot(t_range, self.wbc_feet_vel[:, i % 3, np.int(i/3)], color='b', linewidth=3, marker='')
plt.plot(t_range, self.planner_vgoals[:, i % 3, np.int(i/3)], color='r', linewidth=3, marker='')
plt.plot(t_range, self.wbc_feet_vel_invkin[:, i % 3, np.int(i/3)],
color='darkviolet', linewidth=3, linestyle="--", marker='')
plt.legend([lgd_Y[i % 3] + " " + lgd_X[np.int(i/3)], lgd_Y[i %
3] + " " + lgd_X[np.int(i/3)]+" Ref"], prop={'size': 8})
plt.suptitle("Measured and Reference feet velocities (world frame)")
lgd_X = ["FL", "FR", "HL", "HR"]
lgd_Y = ["Acc X", "Acc Y", "Acc Z"]
plt.figure()
for i in range(12):
if i == 0:
ax0 = plt.subplot(3, 4, index12[i])
else:
plt.subplot(3, 4, index12[i], sharex=ax0)
plt.plot(t_range, self.planner_agoals[:, i % 3, np.int(i/3)], color='r', linewidth=3, marker='')
plt.legend([lgd_Y[i % 3] + " " + lgd_X[np.int(i/3)]+" Ref"], prop={'size': 8})
plt.suptitle("Reference feet accelerations (world frame)")
# LOG_Q
lgd = ["Position X", "Position Y", "Position Z", "Position Roll", "Position Pitch", "Position Yaw"]
plt.figure()
for i in range(6):
if i == 0:
ax0 = plt.subplot(3, 2, index6[i])
else:
plt.subplot(3, 2, index6[i], sharex=ax0)
plt.plot(t_range, self.planner_xref[:, i, 0], "b", linewidth=2)
plt.plot(t_range, self.planner_xref[:, i, 1], "r", linewidth=3)
if i < 3:
plt.plot(t_range, loggerSensors.mocapPosition[:, i], "k", linewidth=3)
else:
plt.plot(t_range, self.mocap_RPY[:, i-3], "k", linewidth=3)
# plt.plot(t_range, self.log_q[i, :], "grey", linewidth=4)
# plt.plot(t_range[:-2], self.log_x_invkin[i, :-2], "g", linewidth=2)
# plt.plot(t_range[:-2], self.log_x_ref_invkin[i, :-2], "violet", linewidth=2, linestyle="--")
Pierre-Alexandre Leziart
committed
plt.legend(["Robot state", "Robot reference state", "Ground truth"], prop={'size': 8})
plt.ylabel(lgd[i])
plt.suptitle("Measured & Reference position and orientation")
# LOG_V
lgd = ["Linear vel X", "Linear vel Y", "Linear vel Z",
"Angular vel Roll", "Angular vel Pitch", "Angular vel Yaw"]
plt.figure()
for i in range(6):
if i == 0:
ax0 = plt.subplot(3, 2, index6[i])
else:
plt.subplot(3, 2, index6[i], sharex=ax0)
plt.plot(t_range, self.esti_v_filt[:, i], "b", linewidth=2)
plt.plot(t_range, self.joy_v_ref[:, i], "r", linewidth=3)
if i < 3:
plt.plot(t_range, self.mocap_b_v[:, i], "k", linewidth=3)
Pierre-Alexandre Leziart
committed
# plt.plot(t_range, self.esti_FK_lin_vel[:, i], "violet", linewidth=3, linestyle="--")
else:
plt.plot(t_range, self.mocap_b_w[:, i-3], "k", linewidth=3)
# plt.plot(t_range, self.log_dq[i, :], "g", linewidth=2)
# plt.plot(t_range[:-2], self.log_dx_invkin[i, :-2], "g", linewidth=2)
# plt.plot(t_range[:-2], self.log_dx_ref_invkin[i, :-2], "violet", linewidth=2, linestyle="--")
Pierre-Alexandre Leziart
committed
plt.legend(["Robot state", "Robot reference state", "Ground truth"], prop={'size': 8})
plt.ylabel(lgd[i])
plt.suptitle("Measured & Reference linear and angular velocities")
"""plt.figure()
plt.plot(t_range[:-2], self.log_x[6, :-2], "b", linewidth=2)
plt.plot(t_range[:-2], self.log_x_cmd[6, :-2], "r", linewidth=2)
plt.plot(t_range[:-2], self.log_dx_invkin[0, :-2], "g", linewidth=2)
plt.plot(t_range[:-2], self.log_dx_ref_invkin[0, :-2], "violet", linewidth=2)
plt.legend(["WBC integrated output state", "Robot reference state",
"Task current state", "Task reference state"])"""
# Analysis of the footstep locations (current and future) with a slider to move along time
# self.slider_predicted_footholds()
# Analysis of the footholds locations during the whole experiment
"""import utils_mpc
import pinocchio as pin
f_c = ["r", "b", "forestgreen", "rebeccapurple"]
quat = np.zeros((4, 1))
steps = np.zeros((12, 1))
o_step = np.zeros((3, 1))
plt.figure()
plt.plot(self.loop_o_q_int[:, 0], self.loop_o_q_int[:, 1], linewidth=2, color="k")
for i in range(self.planner_fsteps.shape[0]):
fsteps = self.planner_fsteps[i]
RPY = utils_mpc.quaternionToRPY(self.loop_o_q_int[i, 3:7])
quat[:, 0] = utils_mpc.EulerToQuaternion([0.0, 0.0, RPY[2]])
oRh = pin.Quaternion(quat).toRotationMatrix()
for j in range(4):
#if np.any(fsteps[k, (j*3):((j+1)*3)]) and not np.array_equal(steps[(j*3):((j+1)*3), 0],
# fsteps[k, (j*3):((j+1)*3)]):
# steps[(j*3):((j+1)*3), 0] = fsteps[k, (j*3):((j+1)*3)]
# o_step[:, 0:1] = oRh @ steps[(j*3):((j+1)*3), 0:1] + self.loop_o_q_int[i:(i+1), 0:3].transpose()
o_step[:, 0:1] = oRh @ fsteps[0:1, (j*3):((j+1)*3)].transpose() + self.loop_o_q_int[i:(i+1), 0:3].transpose()
plt.plot(o_step[0, 0], o_step[1, 0], linestyle=None, linewidth=1, marker="o", color=f_c[j])
lgd1 = ["HAA", "HFE", "Knee"]
lgd2 = ["FL", "FR", "HL", "HR"]
plt.figure()
for i in range(12):
if i == 0:
ax0 = plt.subplot(3, 4, index12[i])
else:
plt.subplot(3, 4, index12[i], sharex=ax0)
tau_fb = self.wbc_P[:, i] * (self.wbc_q_des[:, i] - self.esti_q_filt[:, 7+i]) + \
self.wbc_D[:, i] * (self.wbc_v_des[:, i] - self.esti_v_filt[:, 6+i])
h1, = plt.plot(t_range, self.wbc_tau_ff[:, i], "r", linewidth=3)
h2, = plt.plot(t_range, tau_fb, "b", linewidth=3)
h3, = plt.plot(t_range, self.wbc_tau_ff[:, i] + tau_fb, "g", linewidth=3)
h4, = plt.plot(t_range[:-1], loggerSensors.torquesFromCurrentMeasurment[1:, i],
"violet", linewidth=3, linestyle="--")
plt.xlabel("Time [s]")
plt.ylabel(lgd1[i % 3]+" "+lgd2[int(i/3)]+" [Nm]")
tmp = lgd1[i % 3]+" "+lgd2[int(i/3)]
plt.legend([h1, h2, h3, h4], ["FF "+tmp, "FB "+tmp, "PD+ "+tmp, "Meas "+tmp], prop={'size': 8})
plt.ylim([-8.0, 8.0])
plt.suptitle("FF torques & FB torques & Sent torques & Meas torques")
lgd1 = ["Ctct force X", "Ctct force Y", "Ctct force Z"]
lgd2 = ["FL", "FR", "HL", "HR"]
plt.figure()
for i in range(12):
if i == 0:
ax0 = plt.subplot(3, 4, index12[i])
else:
plt.subplot(3, 4, index12[i], sharex=ax0)
h1, = plt.plot(t_range, self.mpc_x_f[:, 12+i, 0], "r", linewidth=3)
h2, = plt.plot(t_range, self.wbc_f_ctc[:, i], "b", linewidth=3, linestyle="--")
plt.xlabel("Time [s]")
plt.ylabel(lgd1[i % 3]+" "+lgd2[int(i/3)]+" [N]")
plt.legend([h1, h2], ["MPC " + lgd1[i % 3]+" "+lgd2[int(i/3)],
"WBC " + lgd1[i % 3]+" "+lgd2[int(i/3)]], prop={'size': 8})
if (i % 3) == 2:
plt.ylim([-0.0, 26.0])
else:
plt.ylim([-26.0, 26.0])
plt.suptitle("Contact forces (MPC command) & WBC QP output")
lgd1 = ["HAA", "HFE", "Knee"]
lgd2 = ["FL", "FR", "HL", "HR"]
plt.figure()
for i in range(12):
if i == 0:
ax0 = plt.subplot(3, 4, index12[i])
else:
plt.subplot(3, 4, index12[i], sharex=ax0)
h1, = plt.plot(t_range, self.wbc_q_des[:, i], color='r', linewidth=3)
h2, = plt.plot(t_range, self.esti_q_filt[:, 7+i], color='b', linewidth=3)
plt.xlabel("Time [s]")
plt.ylabel(lgd1[i % 3]+" "+lgd2[int(i/3)]+" [rad]")
plt.legend([h1, h2], ["Ref "+lgd1[i % 3]+" "+lgd2[int(i/3)],
lgd1[i % 3]+" "+lgd2[int(i/3)]], prop={'size': 8})
plt.suptitle("Desired actuator positions & Measured actuator positions")
# Evolution of predicted trajectory along time
log_t_pred = np.array([k*self.dt*10 for k in range(self.mpc_x_f.shape[2])])
log_t_ref = np.array([k*self.dt*10 for k in range(self.planner_xref.shape[2])])
"""from IPython import embed
embed()"""
titles = ["X", "Y", "Z", "Roll", "Pitch", "Yaw"]
Pierre-Alexandre Leziart
committed
step = 1000
plt.figure()
for j in range(6):
plt.subplot(3, 2, index6[j])
c = [[i/(self.mpc_x_f.shape[0]+5), 0.0, i/(self.mpc_x_f.shape[0]+5)]
for i in range(0, self.mpc_x_f.shape[0], step)]
for i in range(0, self.mpc_x_f.shape[0], step):
h1, = plt.plot(log_t_pred+(i+10)*self.dt,
self.mpc_x_f[i, j, :], "b", linewidth=2, color=c[int(i/step)])
h2, = plt.plot(log_t_ref+i*self.dt,
self.planner_xref[i, j, :], linestyle="--", marker='x', color="g", linewidth=2)
Pierre-Alexandre Leziart
committed
#h3, = plt.plot(np.array([k*self.dt for k in range(self.mpc_x_f.shape[0])]),
# self.planner_xref[:, j, 0], linestyle=None, marker='x', color="r", linewidth=1)
plt.xlabel("Time [s]")
plt.legend([h1, h2, h3], ["Output trajectory of MPC",
Pierre-Alexandre Leziart
committed
"Input trajectory of planner"]) #, "Actual robot trajectory"])
plt.title("Predicted trajectory for " + titles[j])
plt.suptitle("Analysis of trajectories in position and orientation computed by the MPC")
plt.figure()
for j in range(6):
plt.subplot(3, 2, index6[j])
c = [[i/(self.mpc_x_f.shape[0]+5), 0.0, i/(self.mpc_x_f.shape[0]+5)]
for i in range(0, self.mpc_x_f.shape[0], step)]
for i in range(0, self.mpc_x_f.shape[0], step):
h1, = plt.plot(log_t_pred+(i+10)*self.dt,
self.mpc_x_f[i, j+6, :], "b", linewidth=2, color=c[int(i/step)])
h2, = plt.plot(log_t_ref+i*self.dt,
self.planner_xref[i, j+6, :], linestyle="--", marker='x', color="g", linewidth=2)
h3, = plt.plot(np.array([k*self.dt for k in range(self.mpc_x_f.shape[0])]),
self.planner_xref[:, j+6, 0], linestyle=None, marker='x', color="r", linewidth=1)
plt.xlabel("Time [s]")
plt.legend([h1, h2, h3], ["Output trajectory of MPC",
"Input trajectory of planner", "Actual robot trajectory"])
plt.title("Predicted trajectory for velocity in " + titles[j])
plt.suptitle("Analysis of trajectories of linear and angular velocities computed by the MPC")
Pierre-Alexandre Leziart
committed
step = 1000
lgd1 = ["Ctct force X", "Ctct force Y", "Ctct force Z"]
lgd2 = ["FL", "FR", "HL", "HR"]
plt.figure()
for i in range(12):
if i == 0:
ax0 = plt.subplot(3, 4, index12[i])
else:
plt.subplot(3, 4, index12[i], sharex=ax0)
h1, = plt.plot(t_range, self.mpc_x_f[:, 12+i, 0], "r", linewidth=3)
h2, = plt.plot(t_range, self.wbc_f_ctc[:, i], "b", linewidth=3, linestyle="--")
plt.xlabel("Time [s]")
plt.ylabel(lgd1[i % 3]+" "+lgd2[int(i/3)]+" [N]")
plt.legend([h1, h2], ["MPC " + lgd1[i % 3]+" "+lgd2[int(i/3)],
"WBC " + lgd1[i % 3]+" "+lgd2[int(i/3)]], prop={'size': 8})
if (i % 3) == 2:
plt.ylim([-0.0, 26.0])
else:
plt.ylim([-26.0, 26.0])
plt.suptitle("Contact forces (MPC command) & WBC QP output")
lgd1 = ["Ctct force X", "Ctct force Y", "Ctct force Z"]
lgd2 = ["FL", "FR", "HL", "HR"]
plt.figure()
Pierre-Alexandre Leziart
committed
for i in range(4):
Pierre-Alexandre Leziart
committed
ax0 = plt.subplot(1, 4, i+1)
Pierre-Alexandre Leziart
committed
plt.subplot(1, 4, i+1, sharex=ax0)
for k in range(0, self.mpc_x_f.shape[0], step):
Pierre-Alexandre Leziart
committed
h2, = plt.plot(log_t_pred+k*self.dt, self.mpc_x_f[k, 12+(3*i+2), :], linestyle="--", marker='x', linewidth=2)
h1, = plt.plot(t_range, self.mpc_x_f[:, 12+(3*i+2), 0], "r", linewidth=3)
# h3, = plt.plot(t_range, self.wbc_f_ctc[:, i], "b", linewidth=3, linestyle="--")
Pierre-Alexandre Leziart
committed
plt.plot(t_range, self.esti_feet_status[:, i], "k", linestyle="--")
plt.xlabel("Time [s]")
Pierre-Alexandre Leziart
committed
plt.ylabel(lgd2[i]+" [N]")
plt.legend([h1, h2], ["MPC "+lgd2[i],
"MPC "+lgd2[i]+" trajectory"])
plt.ylim([-1.0, 26.0])
plt.suptitle("Contact forces trajectories & Actual forces trajectories")
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
# Analysis of the complementary filter behaviour
clr = ["b", "darkred", "forestgreen"]
# Velocity complementary filter
lgd_Y = ["dx", "ddx", "alpha dx", "dx_out", "dy", "ddy", "alpha dy", "dy_out", "dz", "ddz", "alpha dz", "dz_out"]
plt.figure()
for i in range(12):
if i == 0:
ax0 = plt.subplot(3, 4, i+1)
else:
plt.subplot(3, 4, i+1, sharex=ax0)
if i % 4 == 0:
plt.plot(t_range, self.esti_HP_x[:, int(i/4)], color=clr[int(i/4)], linewidth=3, marker='') # x input of the velocity complementary filter
elif i % 4 == 1:
plt.plot(t_range, self.esti_HP_dx[:, int(i/4)], color=clr[int(i/4)], linewidth=3, marker='') # dx input of the velocity complementary filter
elif i % 4 == 2:
plt.plot(t_range, self.esti_HP_alpha[:, int(i/4)], color=clr[int(i/4)], linewidth=3, marker='') # alpha parameter of the velocity complementary filter
else:
plt.plot(t_range, self.esti_HP_filt_x[:, int(i/4)], color=clr[int(i/4)], linewidth=3, marker='') # filtered output of the velocity complementary filter
plt.legend([lgd_Y[i]], prop={'size': 8})
plt.suptitle("Evolution of the quantities of the velocity complementary filter")
# Position complementary filter
lgd_Y = ["x", "dx", "alpha x", "x_out", "y", "dy", "alpha y", "y_out", "z", "dz", "alpha z", "z_out"]
plt.figure()
for i in range(12):
if i == 0:
ax0 = plt.subplot(3, 4, i+1)
else:
plt.subplot(3, 4, i+1, sharex=ax0)
if i % 4 == 0:
plt.plot(t_range, self.esti_LP_x[:, int(i/4)], color=clr[int(i/4)], linewidth=3, marker='') # x input of the position complementary filter
elif i % 4 == 1:
plt.plot(t_range, self.esti_LP_dx[:, int(i/4)], color=clr[int(i/4)], linewidth=3, marker='') # dx input of the position complementary filter
elif i % 4 == 2:
plt.plot(t_range, self.esti_LP_alpha[:, int(i/4)], color=clr[int(i/4)], linewidth=3, marker='') # alpha parameter of the position complementary filter
else:
plt.plot(t_range, self.esti_LP_filt_x[:, int(i/4)], color=clr[int(i/4)], linewidth=3, marker='') # filtered output of the position complementary filter
plt.legend([lgd_Y[i]], prop={'size': 8})
plt.suptitle("Evolution of the quantities of the position complementary filter")
plt.show(block=True)
from IPython import embed
embed()
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
def saveAll(self, loggerSensors, fileName="data"):
date_str = datetime.now().strftime('_%Y_%m_%d_%H_%M')
np.savez(fileName + date_str + ".npz",
joy_v_ref=self.joy_v_ref,
esti_feet_status=self.esti_feet_status,
esti_feet_goals=self.esti_feet_goals,
esti_q_filt=self.esti_q_filt,
esti_v_filt=self.esti_v_filt,
esti_v_secu=self.esti_v_secu,
esti_FK_lin_vel=self.esti_FK_lin_vel,
esti_FK_xyz=self.esti_FK_xyz,
esti_xyz_mean_feet=self.esti_xyz_mean_feet,
esti_HP_x=self.esti_HP_x,
esti_HP_dx=self.esti_HP_dx,
esti_HP_alpha=self.esti_HP_alpha,
esti_HP_filt_x=self.esti_HP_filt_x,
esti_LP_x=self.esti_LP_x,
esti_LP_dx=self.esti_LP_dx,
esti_LP_alpha=self.esti_LP_alpha,
esti_LP_filt_x=self.esti_LP_filt_x,
Pierre-Alexandre Leziart
committed
esti_kf_X=self.esti_kf_X,
esti_kf_Z=self.esti_kf_Z,
loop_o_q_int=self.loop_o_q_int,
loop_o_v=self.loop_o_v,
Pierre-Alexandre Leziart
committed
planner_q_static=self.planner_q_static,
planner_RPY_static=self.planner_RPY_static,
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
planner_xref=self.planner_xref,
planner_fsteps=self.planner_fsteps,
planner_gait=self.planner_gait,
planner_goals=self.planner_goals,
planner_vgoals=self.planner_vgoals,
planner_agoals=self.planner_agoals,
planner_is_static=self.planner_is_static,
planner_h_ref=self.planner_h_ref,
mpc_x_f=self.mpc_x_f,
wbc_x_f=self.wbc_x_f,
wbc_P=self.wbc_P,
wbc_D=self.wbc_D,
wbc_q_des=self.wbc_q_des,
wbc_v_des=self.wbc_v_des,
wbc_tau_ff=self.wbc_tau_ff,
wbc_f_ctc=self.wbc_f_ctc,
wbc_feet_pos=self.wbc_feet_pos,
wbc_feet_err=self.wbc_feet_err,
wbc_feet_vel=self.wbc_feet_vel,
tstamps=self.tstamps,
q_mes=loggerSensors.q_mes,
v_mes=loggerSensors.v_mes,
baseOrientation=loggerSensors.baseOrientation,
baseAngularVelocity=loggerSensors.baseAngularVelocity,
baseLinearAcceleration=loggerSensors.baseLinearAcceleration,
baseAccelerometer=loggerSensors.baseAccelerometer,
torquesFromCurrentMeasurment=loggerSensors.torquesFromCurrentMeasurment,
mocapPosition=loggerSensors.mocapPosition,
mocapVelocity=loggerSensors.mocapVelocity,
mocapAngularVelocity=loggerSensors.mocapAngularVelocity,
mocapOrientationMat9=loggerSensors.mocapOrientationMat9,
mocapOrientationQuat=loggerSensors.mocapOrientationQuat,
)
Pierre-Alexandre Leziart
committed
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
def loadAll(self, loggerSensors, fileName=None):
if fileName is None:
import glob
fileName = np.sort(glob.glob('data_2021_*.npz'))[-1] # Most recent file
data = np.load(fileName)
# Load LoggerControl arrays
self.joy_v_ref = data["joy_v_ref"]
self.logSize = self.joy_v_ref.shape[0]
self.esti_feet_status = data["esti_feet_status"]
self.esti_feet_goals = data["esti_feet_goals"]
self.esti_q_filt = data["esti_q_filt"]
self.esti_v_filt = data["esti_v_filt"]
self.esti_v_secu = data["esti_v_secu"]
self.esti_FK_lin_vel = data["esti_FK_lin_vel"]
self.esti_FK_xyz = data["esti_FK_xyz"]
self.esti_xyz_mean_feet = data["esti_xyz_mean_feet"]
self.esti_HP_x = data["esti_HP_x"]
self.esti_HP_dx = data["esti_HP_dx"]
self.esti_HP_alpha = data["esti_HP_alpha"]
self.esti_HP_filt_x = data["esti_HP_filt_x"]
self.esti_LP_x = data["esti_LP_x"]
self.esti_LP_dx = data["esti_LP_dx"]
self.esti_LP_alpha = data["esti_LP_alpha"]
self.esti_LP_filt_x = data["esti_LP_filt_x"]
self.esti_kf_X = data["esti_kf_X"]
self.esti_kf_Z = data["esti_kf_Z"]
self.loop_o_q_int = data["loop_o_q_int"]
self.loop_o_v = data["loop_o_v"]
self.planner_q_static = data["planner_q_static"]
self.planner_RPY_static = data["planner_RPY_static"]
self.planner_xref = data["planner_xref"]
self.planner_fsteps = data["planner_fsteps"]
self.planner_gait = data["planner_gait"]
self.planner_goals = data["planner_goals"]
self.planner_vgoals = data["planner_vgoals"]
self.planner_agoals = data["planner_agoals"]
self.planner_is_static = data["planner_is_static"]
self.planner_h_ref = data["planner_h_ref"]
self.mpc_x_f = data["mpc_x_f"]
self.wbc_x_f = data["wbc_x_f"]
self.wbc_P = data["wbc_P"]
self.wbc_D = data["wbc_D"]
self.wbc_q_des = data["wbc_q_des"]
self.wbc_v_des = data["wbc_v_des"]
self.wbc_tau_ff = data["wbc_tau_ff"]
self.wbc_f_ctc = data["wbc_f_ctc"]
self.wbc_feet_pos = data["wbc_feet_pos"]
self.wbc_feet_err = data["wbc_feet_err"]
self.wbc_feet_vel = data["wbc_feet_vel"]
self.tstamps = data["tstamps"]
# Load LoggerSensors arrays
loggerSensors.q_mes = data["q_mes"]
loggerSensors.v_mes = data["v_mes"]
loggerSensors.baseOrientation = data["baseOrientation"]
loggerSensors.baseAngularVelocity = data["baseAngularVelocity"]
loggerSensors.baseLinearAcceleration = data["baseLinearAcceleration"]
loggerSensors.baseAccelerometer = data["baseAccelerometer"]
loggerSensors.torquesFromCurrentMeasurment = data["torquesFromCurrentMeasurment"]
loggerSensors.mocapPosition = data["mocapPosition"]
loggerSensors.mocapVelocity = data["mocapVelocity"]
loggerSensors.mocapAngularVelocity = data["mocapAngularVelocity"]
loggerSensors.mocapOrientationMat9 = data["mocapOrientationMat9"]
loggerSensors.mocapOrientationQuat = data["mocapOrientationQuat"]
loggerSensors.logSize = loggerSensors.q_mes.shape[0]
def slider_predicted_trajectory(self):
from matplotlib import pyplot as plt
from matplotlib.widgets import Slider, Button
# The parametrized function to be plotted
def f(t, time):
return np.sin(2 * np.pi * t) + time
index6 = [1, 3, 5, 2, 4, 6]
log_t_pred = np.array([(k+1)*self.dt*10 for k in range(self.mpc_x_f.shape[2])])
log_t_ref = np.array([k*self.dt*10 for k in range(self.planner_xref.shape[2])])
trange = np.max([np.max(log_t_pred), np.max(log_t_ref)])
h1s = []
h2s = []
axs = []
h1s_vel = []
h2s_vel = []
axs_vel = []
# Define initial parameters
init_time = 0.0
# Create the figure and the line that we will manipulate
fig = plt.figure()
ax = plt.gca()
for j in range(6):
ax = plt.subplot(3, 2, index6[j])
h1, = plt.plot(log_t_pred, self.mpc_x_f[0, j, :], "b", linewidth=2)
h2, = plt.plot(log_t_ref, self.planner_xref[0, j, :], linestyle="--", marker='x', color="g", linewidth=2)
h3, = plt.plot(np.array([k*self.dt for k in range(self.mpc_x_f.shape[0])]),
self.planner_xref[:, j, 0], linestyle=None, marker='x', color="r", linewidth=1)
Pierre-Alexandre Leziart
committed
axs.append(ax)
h1s.append(h1)
h2s.append(h2)
#ax.set_xlabel('Time [s]')
axcolor = 'lightgoldenrodyellow'
#ax.margins(x=0)
# Make a horizontal slider to control the time.
axtime = plt.axes([0.25, 0.03, 0.65, 0.03], facecolor=axcolor)
time_slider = Slider(
ax=axtime,
label='Time [s]',
valmin=0.0,
valmax=self.logSize*self.dt,
valinit=init_time,
)
# Create the figure and the line that we will manipulate (for velocities)
fig_vel = plt.figure()
ax = plt.gca()
for j in range(6):
ax = plt.subplot(3, 2, index6[j])
h1, = plt.plot(log_t_pred, self.mpc_x_f[0, j, :], "b", linewidth=2)
h2, = plt.plot(log_t_ref, self.planner_xref[0, j, :], linestyle="--", marker='x', color="g", linewidth=2)
h3, = plt.plot(np.array([k*self.dt for k in range(self.mpc_x_f.shape[0])]),
self.planner_xref[:, j+6, 0], linestyle=None, marker='x', color="r", linewidth=1)
Pierre-Alexandre Leziart
committed
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
axs_vel.append(ax)
h1s_vel.append(h1)
h2s_vel.append(h2)
#axcolor = 'lightgoldenrodyellow'
#ax.margins(x=0)
# Make a horizontal slider to control the time.
axtime_vel = plt.axes([0.25, 0.03, 0.65, 0.03], facecolor=axcolor)
time_slider_vel = Slider(
ax=axtime_vel,
label='Time [s]',
valmin=0.0,
valmax=self.logSize*self.dt,
valinit=init_time,
)
# The function to be called anytime a slider's value changes
def update(val, recursive=False):
time_slider.val = np.round(val / (self.dt*10), decimals=0) * (self.dt*10)
rounded = int(np.round(time_slider.val / self.dt, decimals=0))
for j in range(6):
h1s[j].set_xdata(log_t_pred + time_slider.val)
h2s[j].set_xdata(log_t_ref + time_slider.val)
y1 = self.mpc_x_f[rounded, j, :] - self.planner_xref[rounded, j, 1:]
y2 = self.planner_xref[rounded, j, :] - self.planner_xref[rounded, j, :]
h1s[j].set_ydata(y1)
h2s[j].set_ydata(y2)
axs[j].set_xlim([time_slider.val - self.dt * 3, time_slider.val+trange+self.dt * 3])
ymin = np.min([np.min(y1), np.min(y2)])
ymax = np.max([np.max(y1), np.max(y2)])
axs[j].set_ylim([ymin - 0.05 * (ymax - ymin), ymax + 0.05 * (ymax - ymin)])
fig.canvas.draw_idle()
if not recursive:
update_vel(time_slider.val, True)
def update_vel(val, recursive=False):
time_slider_vel.val = np.round(val / (self.dt*10), decimals=0) * (self.dt*10)
rounded = int(np.round(time_slider_vel.val / self.dt, decimals=0))
for j in range(6):
h1s_vel[j].set_xdata(log_t_pred + time_slider.val)
h2s_vel[j].set_xdata(log_t_ref + time_slider.val)
y1 = self.mpc_x_f[rounded, j+6, :]
y2 = self.planner_xref[rounded, j+6, :]
h1s_vel[j].set_ydata(y1)
h2s_vel[j].set_ydata(y2)
axs_vel[j].set_xlim([time_slider.val - self.dt * 3, time_slider.val+trange+self.dt * 3])
ymin = np.min([np.min(y1), np.min(y2)])
ymax = np.max([np.max(y1), np.max(y2)])
axs_vel[j].set_ylim([ymin - 0.05 * (ymax - ymin), ymax + 0.05 * (ymax - ymin)])
fig_vel.canvas.draw_idle()
if not recursive:
update(time_slider_vel.val, True)
# register the update function with each slider
time_slider.on_changed(update)
time_slider_vel.on_changed(update)
plt.show()
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
def slider_predicted_footholds(self):
from matplotlib import pyplot as plt
from matplotlib.widgets import Slider, Button
import utils_mpc
import pinocchio as pin
self.planner_fsteps
# Define initial parameters
init_time = 0.0
# Create the figure and the line that we will manipulate
fig = plt.figure()
ax = plt.gca()
h1s = []
f_c = ["r", "b", "forestgreen", "rebeccapurple"]
quat = np.zeros((4, 1))
fsteps = self.planner_fsteps[0]
o_step = np.zeros((3*int(fsteps.shape[0]), 1))
RPY = utils_mpc.quaternionToRPY(self.loop_o_q_int[0, 3:7])
quat[:, 0] = utils_mpc.EulerToQuaternion([0.0, 0.0, RPY[2]])
oRh = pin.Quaternion(quat).toRotationMatrix()
for j in range(4):
o_step[0:3, 0:1] = oRh @ fsteps[0:1, (j*3):((j+1)*3)].transpose() + self.loop_o_q_int[0:1, 0:3].transpose()
h1, = plt.plot(o_step[0::3, 0], o_step[1::3, 0], linestyle=None, linewidth=0, marker="o", color=f_c[j])
h1s.append(h1)
axcolor = 'lightgoldenrodyellow'
# Make a horizontal slider to control the time.
axtime = plt.axes([0.25, 0.03, 0.65, 0.03], facecolor=axcolor)
time_slider = Slider(
ax=axtime,
label='Time [s]',
valmin=0.0,
valmax=self.logSize*self.dt,
valinit=init_time,
)
ax.set_xlim([-0.3, 0.5])
ax.set_ylim([-0.3, 0.5])
# The function to be called anytime a slider's value changes
def update(val):
time_slider.val = np.round(val / (self.dt*10), decimals=0) * (self.dt*10)
rounded = int(np.round(time_slider.val / self.dt, decimals=0))
fsteps = self.planner_fsteps[rounded]
o_step = np.zeros((3*int(fsteps.shape[0]), 1))
RPY = utils_mpc.quaternionToRPY(self.loop_o_q_int[rounded, 3:7])
quat[:, 0] = utils_mpc.EulerToQuaternion([0.0, 0.0, RPY[2]])
oRh = pin.Quaternion(quat).toRotationMatrix()
for j in range(4):
for k in range(int(fsteps.shape[0])):
o_step[(3*k):(3*(k+1)), 0:1] = oRh @ fsteps[(k):(k+1), (j*3):((j+1)*3)].transpose() + self.loop_o_q_int[rounded:(rounded+1), 0:3].transpose()
h1s[j].set_xdata(o_step[0::3, 0].copy())
h1s[j].set_ydata(o_step[1::3, 0].copy())
fig.canvas.draw_idle()
# register the update function with each slider
time_slider.on_changed(update)
plt.show()
Pierre-Alexandre Leziart
committed
if __name__ == "__main__":
import LoggerSensors
# Create loggers
loggerSensors = LoggerSensors.LoggerSensors(logSize=5997)
logger = LoggerControl(0.002, 100, logSize=5997)
# Load data from .npz file
logger.loadAll(loggerSensors)
# Call all ploting functions
#logger.plotAll(loggerSensors)
logger.slider_predicted_trajectory()