Whole-Body End-Effector Pose Tracking

Tifanny Portela Andrei Cramariuc Mayank Mittal Marco Hutter

International Conference on Robotics and Automation (ICRA) 2025

Combining manipulation with the mobility of legged robots is essential for a wide range of robotic applications. However, integrating an arm with a mobile base significantly increases the system's complexity, making precise end-effector control challenging. Existing model-based approaches are often constrained by their modeling assumptions, leading to limited robustness. Meanwhile, recent Reinforcement Learning (RL) implementations restrict the arm's workspace to be in front of the robot or track only the position to obtain decent tracking accuracy. In this work, we address these limitations by introducing a whole-body RL formulation for end-effector pose tracking in a large workspace on rough terrains. Our proposed method involves a terrain-aware sampling strategy for the robot's initial configuration and end-effector pose commands, as well as a game-based curriculum to extend the robot's operating range. We validate our approach on the ANYmal quadrupedal robot with a six DoF robotic arm. Through our experiments, we show that the learned controller achieves precise command tracking over a large workspace and adapts across varying terrains such as stairs and slopes. On deployment, it achieves a pose-tracking error of 2.64 cm and 3.64 degrees, outperforming existing competitive baselines.

We train an RL controller to track end-effector pose commands in simulation and transfer to a mobile legged manipulator.

End-effector pose tracking evaluation, using a motion capture system, on flat terrain across 20 randomly sampled poses

We evaluate the tracking performance of our whole-body RL controller on stairs. The average position error reaches 2.64 cm, and the average orientation error 3.64 degrees.

End-effector pose tracking evaluation, using a motion capture system, on stairs across 20 randomly sampled poses

The system can handle unmodeled added weight on the end-effector, allowing for the attachment of various end-effectors and dynamic carrying of unknown payloads during operation.

Up to 3.75 kg when stationary

Up to 1.3 kg when in movement

We compare the tracking performance of our learned whole-body RL controller to a model-based MPC baseline on flat terrain. While both perform similarly in median accuracy, the RL policy achieves significantly lower mean errors by avoiding self-collisions and local minimas during transitions between distant poses.


          @inproceedings{portela2025whole,
            title={Whole-body end-effector pose tracking},
            author={Portela, Tifanny and Cramariuc, Andrei and Mittal, Mayank and Hutter, Marco},
            booktitle={2025 IEEE International Conference on Robotics and Automation (ICRA)},
            year={2025},
            organization={IEEE}
          }