Scaling Whole-body Multi-contact Manipulation with Contact Optimization

TL;DR

This work tackles the scalability of whole-body manipulation planning for humanoid robots by introducing a differentiable, closed-form surface proximity representation that supports continuous optimization over contact locations. It couples this representation with a hybrid cost design that blends robot-centric and object-centric manipulability to guide contact selection, enabling autonomous planning across body parts and contact modes. A two-stage pipeline, consisting of a global planner and a trajectory optimizer, demonstrates substantial improvements in planning time (about 77% faster on average) and successful hardware transfer on a humanoid platform. While demonstrated in planar settings, the approach lays groundwork for extending to 3D and, ultimately, real-time control frameworks like MPC, offering a practical route toward efficient, contact-rich humanoid manipulation.

Abstract

Daily tasks require us to use our whole body to manipulate objects, for instance when our hands are unavailable. We consider the issue of providing humanoid robots with the ability to autonomously perform similar whole-body manipulation tasks. In this context, the infinite possibilities for where and how contact can occur on the robot and object surfaces hinder the scalability of existing planning methods, which predominantly rely on discrete sampling. Given the continuous nature of contact surfaces, gradient-based optimization offers a more suitable approach for finding solutions. However, a key remaining challenge is the lack of an efficient representation of robot surfaces. In this work, we propose (i) a representation of robot and object surfaces that enables closed-form computation of proximity points, and (ii) a cost design that effectively guides whole-body manipulation planning. Our experiments demonstrate that the proposed framework can solve problems unaddressed by existing methods, and achieves a 77% improvement in planning time over the state of the art. We also validate the suitability of our approach on real hardware through the whole-body manipulation of boxes by a humanoid robot.

Figures (9)

• Figure 1: enables versatile handling of objects with the full surface of robot body.
• Figure 2: Overview of the planning problem.
• Figure 3: Point Proximity
• Figure 4: Segment Proximity
• Figure 6: Contact surface representation