The Distributed Autonomy lab makes this contribution through a specific focus on search and safety applications through intelligent sensor placement. Securing potential hot spots involves manually operated cameras and human patrols. However, humans are expensive and video feeds are not always adequately monitored. Ultimately, autonomous algorithms will be able to place sensors such as cameras and explosive sensors in positions where they are most likely to identify dangerous or threatening situations. This advance will reduce human exposure to danger and reduce the amount data by focusing attention on specific threats. The underlying science of these problems concerns low-bandwidth, cooperative paradigms. If communications requirements scale with the size of the team but the communications channel is bounded, team sizes are artificially constrained. Our hypotheses center around optimizing message content and targets maximizing performance given limited bandwidth.
Previous approaches rely on brute-force or globally optimal methods. However, bio-inspired methods are able to provide a fresh look at old problems. Such approaches have included individual, rather than global goal selection, emergent consensus based on pre-agreement, and heuristic behaviors. Cooperative paradigms that cooperate based on shared state rather than shared goals or explicated coordinated actions have been shown to provide comparable coverage at smaller team size but better scalability at larger team sizes through simulations and robot experiments. Through collaborations with other faculty members, assumptions about network connectivity are being challenged. As robots move in and out of communications range, we need new protocols that are tolerant of momentary and persistent network failures.
As new students engage in this research, we see opportunities to make robotics more accessible to a new generation of scientists. Our approach uses an open source simulator that also provides an abstraction layer to program a variety of robots. This approach creates a more gradual learning curve to introduce students to sensors and hardware programming. Other research-related projects investigate language and debugging tools for autonomous controller development.
The ultimate goal of the Distributed Autonomy lab is to affect the larger community both through our scientific advances and the new scientists that we produce. The answers to the questions that we investigate provide a mechanism to save lives.
Active Projects: ARTSI Workshop in a Box | Facilitating Operator Interaction with Quality of Service-based Multirobot Surveillance Systems | EMT-Primate-inspired Heterogeneous Mobile and Static Sensor Networks | Real time Java in Behavioral Robotics | ARTSI Resources | Social Networks in Education | PREOP-First-year computer science education (PREOP) | Cyber Physical Systems (CPS) | 2011 SIGCSE Hoedown | Robot Device Interface Specification | UAV Obstacle Avoidance
ClassProjects: Class Projects