Learning Accurate End-Effector Trajectory Tracking with Inertial Feedback for a Quadruped Manipulator
Authors: Dohyeon Kim, Seungyong Han, Yeongtae Jung
Organizations: Graduate School of Mechanical-Aerospace-Electric Convergence Engineering, Jeonbuk National University, Jeollabuk-do 54896, Republic of Korea. · Department of Mechanical System Engineering, Jeonbuk National University, Jeollabuk-do 54896, Republic of Korea. · Advanced Transportation Machinery Research Center, Jeonbuk National University, Jeollabuk-do 54896, Republic of Korea.
Accurate end-effector trajectory tracking on quadruped manipulators is challenging because uncertainties in actuators, foot-ground contact, and geometric errors are propagated through the leg-arm kinematic chain. Detecting such errors requires sensing beyond joint encoders that can capture the effects of these uncertainties on the robot's motion, since joint encoders only provide measurements at the actuated joints and cannot fully observe mechanical errors such as backlash, dimensional inaccuracies, or structural deformation. This work addresses these challenges using a reinforcement learning framework with end-effector (EE) inertial feedback, targeting local high-accuracy manipulation while maintaining a nominal body pose and support configuration without intentional body motion or foot repositioning. The learned policy uses EE inertial measurements to compensate for tracking errors induced by uncertainties throughout the leg-arm system that cannot be fully inferred from proprioception alone. Simulation and real-world experiments on continuous EE trajectory tracking show overall improvements relative to nominal and proprioception-only baselines. In real-world experiments, the proposed policy reduces the MoCap-based RMSE by 17.3% and 28.0% relative to the nominal baseline under the two in-distribution (ID) command conditions, respectively, and by an additional 12.6% and 15.9% relative to the proprioception-only baseline. Furthermore, the trajectory-wise RMSE between FK-reconstructed and motion-capture-measured end-effector positions averaged 12.61+-2.15mm (mean+-standard deviation) across 24 controller-trajectory combinations, revealing a substantial discrepancy between nominal kinematic reconstruction and measured end-effector motion.
Figures & tables
Fig. 1: Overview of the proposed control framework. Nominal leg and arm controllers generate baseline commands, while an arm-agent residual RL policy uses proprioceptive and end-effector inertial feedback to compensate for propagated tracking errors. Sim-to-real alignment combines simulation parameter identification and neural residual modeling of IMU discrepancies.
Fig. 2: Sim-to-real alignment results for the closed-loop arm dynamics and end-effector IMU measurements. (a) Joint-space trajectory matching after simulation parameter identification using chirp excitation. (b) Frequency-domain comparison of end-effector IMU signals before and after applying the neural residual model.
Case
Cmd.
Method
RMSE norm (mm)
MAE norm (mm)
(Vmax, Amax)
Val.
vs Nom. (%)
vs w/o (%)
Val.
vs Nom. (%)
vs w/o (%)
1
ID (0.05, 0.5)
Nominal
6.34
–
–
5.83
–
–
w/o EE IMU
4.09
35.5
–
3.82
34.5
–
w/ EE IMU
3.40
46.4
16.8
3.20
45.1
16.2
2
ID (0.10, 3.0)
Nominal
8.05
–
–
7.70
–
–
w/o EE IMU
6.83
15.2
–
6.60
14.3
–
TABLE I: Simulation Results: End-Effector Position Tracking Error
Fig. 3: Experimental setup consisting of a Unitree Go2 quadruped, a mounted PiPER arm, an end-effector-mounted IMU, and motion-capture markers on the end-effector and the Go2 base.
Case
Cmd.
Method
RMSE norm (mm)
MAE norm (mm)
(Vmax, Amax)
FK
MoCap
vs Nom (%)
vs w/o (%)
FK
MoCap
vs Nom (%)
vs w/o (%)
1
ID (0.05, 0.5)
Nominal
6.67
13.92
–
–
6.15
13.29
–
–
w/o EE IMU
6.40
13.18
5.3
–
6.06
12.57
5.4
–
w/ EE IMU
5.86
11.51
17.3
12.6
5.53
10.83
18.5
13.8
2
ID (0.10, 3.0)
Nominal
11.58
15.06
–
–
10.83
13.62
–
–
w/o EE IMU
10.23
12.90
14.4
–
9.80
11.84
13.1
–
TABLE II: Experimental Results: End-Effector Position Tracking Error
Robot trajectory tracking develops methods that make robots follow desired paths or motions accurately despite disturbances, uncertainty, and physical constraints.
Legged robots equipped with manipulators offer promising capabilities for loco-manipulation tasks in unstructured environments. However, achieving task-consistent end-effector behavior during locomotion remains challenging due to terrain-induced disturbances and the strong coupling between locomotion and manipulation, particularly for small-scale quadruped platforms. A reinforcement-learning based whole-body control framework is proposed for quadruped loco-manipulation on complex terrain. The proposed framework combines a hybrid task representation with task-prioritized whole-body learning to coordinate locomotion and manipulation while preserving task-consistent end-effector behavior. A lightweight quadruped manipulator platform is further developed to validate the proposed framework on a small-scale quadruped system. Real-world experiments demonstrate successful deployment on complex terrains such as slopes up to 30 degrees, stairs, and deformable terrain. These results demonstrate the feasibility of deploying learned whole-body control policy for quadruped loco-manipulation in real-world environments and highlight the importance of task-consistent whole-body coordination for terrain-adaptive loco-manipulation.
Yanzhou Jin, Fei Suo, Chang Liu +5
Graduate School of Engineering, The University of Tokyo, Japan. · Department of Micro-Nano Mechanical Science and Engineering, Nagoya University, Nagoya, 464-8603, Japan. · Department of Mathematics and Informatics, Ecole Centrale de Lyon, France.
Precise end-effector tracking during humanoid whole-body motion is challenging due to floating-base oscillations, gravity, dynamic coupling, and locomotion-induced disturbances. We propose ResGAC, a whole-body humanoid controller for precise end-effector pose tracking that combines geometric admittance control (GAC) with residual reinforcement learning. GAC provides structured \SE task-space feedback and generates nominal arm joint-position targets, while residual RL compensates for unmodeled dynamics and coordinates locomotion and balance in the shared joint-position action space. The left-invariant geometric formulation allows the same GAC law to be used across manipulation reference frames. This enables the use of a ground-attached heading frame that preserves planar locomotion while removing pelvis roll, pitch, and heave from the manipulation reference, thereby reducing reference-induced end-effector motion during locomotion. ResGAC is validated on a real Unitree G1 humanoid. Across four standing end-effector tracking benchmarks, ResGAC consistently outperforms representative baselines, including SONIC, achieving lower translational and rotational errors. Real-world experiments further demonstrate reduced propagation of pelvis motion to the desired end-effector pose using the proposed ground-attached heading frame. ResGAC achieves 90% success in a standing peg-in-hole task compared with 50% for SONIC, and accurate world-frame \SE end-effector pose tracking during lower-body motion. Experimental videos are included in the supplementary material and are also available on the project website: https://resgac.github.io/ResGAC-website/.
Joohwan Seo, Xiaofeng Guo, Jinkun Cao +4
University of California, Berkeley · Amazon FAR · Carnegie Mellon University
Bipedal loco-manipulation enables robots to interact with objects beyond the nominal workspace of their arms by coordinating locomotion and manipulation. Realizing this capability requires a low-level whole-body controller that translates task-level manipulation goals into coordinated arm and leg motions while maintaining balance. We present a unified whole-body controller trained with reinforcement learning that directly maps 6-DoF end-effector targets to coordinated actions for the bipedal base and robotic arm. Given only an end-effector target, the learned controller autonomously coordinates reaching, postural adaptation, and stepping without explicit base-velocity or footstep commands. A reward-gating strategy regulates the trade-offs among end-effector tracking, locomotion, and balance during training, while a temporal context estimator combines windowed Transformer encoding, recurrent GRU memory, and auxiliary dynamics prediction to extract dynamics-relevant information from observation history. Real-robot experiments demonstrate that the same controller supports reaching, postural adaptation, and stepping under commands from VR teleoperation, a learned diffusion policy, and scripted trajectories, providing a common end-effector interface for diverse manipulation tasks.