Whole-Body Control With Terrain Estimation of A 6-DoF Wheeled Bipedal Robot

TL;DR

This work presents a complete 3D dynamics model for a 6-DoF wheeled bipedal robot and a terrain-aware whole-body control framework. Terrain perception is delivered online via LiDAR-based odometry (FAST-LIO) and an improved PCA method to estimate the ground normal, enabling active adaptation to uneven terrain. The control stack comprises estimation, task-space PD/LQR-based task controllers, and a hierarchical optimization that maps desired task accelerations to joint torques while respecting dynamics and contact constraints. Real-world and simulation experiments demonstrate disturbance rejection, height/adaptation on uneven surfaces, and significant gains in terrain handling when ground normals are accurately estimated. The approach holds promise for robust exploration and inspection tasks in unstructured environments, with future work extending to terrain-informed planning and control.

Abstract

Wheeled bipedal robots have garnered increasing attention in exploration and inspection. However, most research simplifies calculations by ignoring leg dynamics, thereby restricting the robot's full motion potential. Additionally, robots face challenges when traversing uneven terrain. To address the aforementioned issue, we develop a complete dynamics model and design a whole-body control framework with terrain estimation for a novel 6 degrees of freedom wheeled bipedal robot. This model incorporates the closed-loop dynamics of the robot and a ground contact model based on the estimated ground normal vector. We use a LiDAR inertial odometry framework and improved Principal Component Analysis for terrain estimation. Task controllers, including PD control law and LQR, are employed for pose control and centroidal dynamics-based balance control, respectively. Furthermore, a hierarchical optimization approach is used to solve the whole-body control problem. We validate the performance of the terrain estimation algorithm and demonstrate the algorithm's robustness and ability to traverse uneven terrain through both simulation and real-world experiments.

Figures (8)

• Figure 1: DIABLO. A 6-DoF wheeled bipedal robot composed entirely of direct drive joints. A Livox Mid-360 LiDAR installed on the robot scans the environment to acquire terrain information. For additional maneuver demos, watch the included video: https://youtu.be/E0AMqLYjEYE.
• Figure 2: Robot coordinate systems, generalized coordinates, actuated torques, and leg structure. (a) $q_{\{left,right\}}$ and $\tau_{\{left,right\}}$ respectively represent the generalized coordinates and the actuated torques on both sides of the robot. $_I\bm{n}_l$ and $_I\bm{n}_r$ are ground normal vectors at the contact points between the wheel and the ground. (b) $B_{\{left,right\}}$ represent the rigid bodies on both sides of the robot. The red line represents a simplified version of the leg linkage. The pendulum angle $\theta$ is the angle between the simplified leg and the vertical direction.
• Figure 3: WBC framework
• Figure 4: Local control frame. The sagittal plane and the horizontal plane of the robot. $C'$ is the orthographic projection of CoM onto the sagittal plane. $C'_r$ is the contact point of the right wheel when the split angle equals zero.
• Figure 5: Disturbance recovery experiment. (a) The robot is disturbed and recovers. (b) The evolution of the robot's pitch angle $\beta$, CoM deviation distance $\bm{r}_{CoM}^x$, absolute position $\bm{s}_{CoM}^x$, and actuated torques $\bm{\tau}_a$.