Autonomous Modification of Unstructured Environments (ICRA'20)

We developed a specialized construction algorithm and robotic system that can autonomously build motion support structures with previously unseen objects. The approach is based on our prior work on adaptive ramp building algorithms, but it eliminates the assumption of having specialized building materials that simplify manipulation and planning for stability. Utilizing irregularly shaped stones makes the problem significantly more challenging since the outcome of individual placements is sensitive to details of contact geometry and friction, which are difficult to observe. To reuse the same high-level algorithm, we develop a new physics-based planner that explicitly considers the uncertainty produced by incomplete in-situ sensing and imprecision during pickup and placement. The resulting system can build structures with more than 70 stones, which in turn provide traversable paths to previously inaccessible locations.

We proposed a method that utilizes simulations to incorporate the uncertainties in the world description and in the robot dynamics. The physics simulator needs to be tuned to favor speed over accuracy to allow for a sufficient number of simulated samples of the world model in reasonable time.

The video shows the 5 experiments that we have done (Video 1). In the video, the robot is deployed on a terrain that is navigable and has access to an abundant supply of stones from a specified region (quarry). The robot system is tasked to build a navigable path to the target area using the stones from the quarry. Starting from the largest stone option, the robot estimates a deposition pose. If no pose was found, it either: (a) moves to the next largest stone until it finds a stone with a valid deposition pose, or (b) requests for more options. Once the building process is completed, the robot finds a navigable path in the structure and attempts to reach the target area. In the experiments, the robot's sensing range is large enough to see the target structure. The presented system is able to build navigable structures over unstructured terrain with found stones without prior knowledge of either the stones or the environment. The system can operate fully autonomously over many hours and find, choose, and deposit stones to build motion support structures.

* This work has accepted to ICRA'20, supported by NSF#184634 and CAPES #013584/2013-08.

** This project was developed while affiliated to the Department of Computer Science & Engineering, at University at Buffalo, The State University of New York.