Home     Research     Publications     News     Talks

For an updated list of cool videos, check the Youtube channel.

Autonomous Modification of Unstructured Environments using Found Material (ICRA 2020)

Adaptive Autonomous Modification of Unstructured Environments

Proposed methods to perform autonomous construction in irregular terrain and designed a heterogeneous robotic system that allows effective evaluation of these methods at a physical implementation level. The heterogeneous robotic team is composed of two autonomous robots that are able to manipulate material of different physical properties (rigid and amorphous materials)

Multi-Robot and Multi-Material (Rigid and Compliante)

Published in: Int. Symp. on Distributed Autonomous Robotic Systems (DARS 2018)

We developed a construction model that incorporates the local information perceived by a robot to define the amount of building material needed to be deposited on the environment to allow for mobility. This model allows the deployment of heterogeneous robot teams to construct structures efficiently by combining materials of different sizes and physical properties, such as rigid and compliant materials. A biological mechanism called Stigmergy is used as a means to coordinate the robots’ construction actions, where information about building actions is encoded in the partially built structure.

Single Robot and Amorphous/Compliate Material

Published in: Autonomous adaptive modification of unstructured environments. Robotics: Science and Systems (RSS 2018)

We use the notion of navigability to enable a robot to modify its environment and provide access to areas which were initially unreachable. In this model, the motion ability of a mobile platform over unstructured terrain is expressed in terms of three scalar parameters that can provide a conservative lower bound of the motion ability based on the stability and clearance constraints of an arbitrary mobile platform. We used a deposition model that provides an upper bound of how much the environment changes in response to its actions. Together, the navigability and deposition model allow the robot to adaptively synthesize low-level motion and manipulation plans that are guaranteed to produce navigable structures as long as the environmental changes produced by the robots are consistent with the deposition model at each step. Navigability parameters induce lower bounds on both the allowable deposition models and sensing abilities of a robot. This abstract model for the construction process directly links the motion, sensing, and manipulation abilities of a robot.

Dry Stacking for Automated Construction with Irregular Objects