Summary Video

Overview

The Stanford/Berkeley Testbed of Autonomous Rotorcraft for Multi-Agent Control (STARMAC) is a multi vehicle testbed used to demonstrate new concepts in multi-agent control on a real-world platform. STARMAC consists of six quadrotor vehicles that are equipped with sufficient sensing and computing power (including a GPS, sonar range-finder, and a Microstrain IMU) to enable completely autonomous operation, from low level tasks (e.g. waypoint and trajectory following) to high-level optimal control strategies (e.g. information theoretic cooperative search and rescue). In order to make such a testbed easy to use, we focused on a small, light, lowcost design, which presented numerous opportunities for innovative work.

The resulting vehicles can fly safely in both indoor and outdoor environments, can carry a wide variety of payloads (including laser range-finders and PC104 computers) and have simple, reconfigurable construction with low maintenance requirements.

Previous work on STARMAC has covered a wide variety of topics, including cooperative search and rescue, information theoretic control, cooperative collision avoidance, optimal path planning, and designing guaranteed safe acrobatic trajectories. Current work is focused on integrating perception capabilities onto STARMAC, to enable the vehicles to understand, respond to, and reason about their environment.

Latest Research

The following are two videos featuring a backflip maneuver.


[ Click here to download in WMV format ]

[ Click here to download in WMV format ]

The purpose of this work was to demonstrate that some of the theories from hybrid systems, and in particular, the analytical tools of reachability, could be applied to a complex real-world system. In essence, reachability can guarantee that a system (i.e. a quadrotor doing a backflip) will be safe (i.e. not crash into the ground) under certain conditions (i.e. windspeeds below some maximum value).

People



<< Back to Hybrid Systems Lab Home



All content on this website ©2009 by the Hybrid Systems Lab
For more information, contact jgillula@eecs.berkeley.edu
Last Updated: 10/15/2009