Sumo: Dynamic and Generalizable Whole-Body Loco-Manipulation
2026-04-09 • Robotics
Robotics
AI summaryⓘ
The authors developed a method that helps legged robots like Spot move and handle big, heavy objects using their whole body. They use a technique that combines a pre-trained control policy with a planner that adjusts actions during the task, allowing the robot to perform different tasks without retraining. Their method works in real life for tasks like flipping a heavy tire and dragging large barriers, and also applies to humanoid robots in simulation. This approach is flexible because it can be adapted simply by changing the task goals on the fly.
legged robotswhole-body controlsim-to-realsample-based plannerloco-manipulationquadruped robotdynamic manipulationtest-time steeringrobot dexteritycontrol policy
Authors
John Z. Zhang, Maks Sorokin, Jan Brüdigam, Brandon Hung, Stephen Phillips, Dmitry Yershov, Farzad Niroui, Tong Zhao, Leonor Fermoselle, Xinghao Zhu, Chao Cao, Duy Ta, Tao Pang, Jiuguang Wang, Preston Culbertson, Zachary Manchester, Simon Le Cléac'h
Abstract
This paper presents a sim-to-real approach that enables legged robots to dynamically manipulate large and heavy objects with whole-body dexterity. Our key insight is that by performing test-time steering of a pre-trained whole-body control policy with a sample-based planner, we can enable these robots to solve a variety of dynamic loco-manipulation tasks. Interestingly, we find our method generalizes to a diverse set of objects and tasks with no additional tuning or training, and can be further enhanced by flexibly adjusting the cost function at test time. We demonstrate the capabilities of our approach through a variety of challenging loco-manipulation tasks on a Spot quadruped robot in the real world, including uprighting a tire heavier than the robot's nominal lifting capacity and dragging a crowd-control barrier larger and taller than the robot itself. Additionally, we show that the same approach can be generalized to humanoid loco-manipulation tasks, such as opening a door and pushing a table, in simulation. Project code and videos are available at \href{https://sumo.rai-inst.com/}{https://sumo.rai-inst.com/}.