Autonomous Airboats
In this project, we are developing a large fleet of small autonomous airboats for applications such as environmental monitoring and flood disaster response. See the latest adventures on the project blog .
Autonomous Vehicle Health Monitoring
In this project, we are using data from our airboats to develop algorithms for looking at sensor calibration errors and possible vehicle failure. Working with SEI and Prof. Jeff Scheider, the project applies machine learning to detect anomolies in sensor data and bring them to the attention of operators. More detail here .
Adjustably Autonomous Robot Teams
In this project, we are creating robot teams that perform complex tasks under the control of a single operator. Our current challenge task is to use robots to manage a car checkpoint, while the operator stays safely removed. A video of our early efforts is here .
Intelligent UAV Interfaces
In this project, sponsored by WPAFB, we are developing intelligent interfaces for UAV operators. The interfaces will provide AI tools, such as path planners, in a way that allows the operator to maintain situational awareness and control.
Robots Dogs
In this project we have two LAGR robots who are shown an object, the object is thrown and the robot has to go and collect it. The robots must be able to distinguish which object it was shown, since there can be other objects in the environment. Our robots will be competing in a competition in April.
Cognitively Compliant Robot Teams
This project, in cooperation with U. Pittsburgh and BYU, is aimed at developing techniques that will allow an operator to have better control over a robot team, by making the team respond more intelligently to operator controls. We hope that new control techniques will emerge as the teams become more intelligent using the input.
Machinetta
Machinetta is our software package that encapsulates coordination algorithms in a proxy . Each robot, person and agent in a team is given a proxy and the proxies cooperate to manage the coordination. Machinetta is an updated, fully distributed, Java version of Teamcore that has been effectively used in a number of domains to coordinate large teams. Download Machinetta from here.
Information Dynamics
In this project, we are looking at the dynamics of information sharing in large teams, connected by social networks. We are looking at both algorithms for more efficient sharing and what happens on a large scale when abstracted information is shared across the network.