The Naval Research Laboratory (NRL) is developing a robot-control system based on the application of genetic algorithms. The approach parameterizes the path-planning algorithm and finds an optimum parameterization based on a tailored metric. The metric is established by simulating the robot in a representative environment and combining relevant measures of performance, such as task success, speed, energy consumption, robot-structure stress, and joint torques among others. Genetic algorithms provide a way to iteratively improve the path-planning parameterization based on repeated evaluation of the metric. Since many simulation executions are required, the use of a supercomputer or computing cluster is needed for fast execution.

Energid Technologies is providing the software for simulating and validating robot designs through its Actin toolkit. Actin provides accessible interfaces, innovative automatic control algorithms, sensor modeling, and dynamic simulation. Dynamic simulation includes both articulation dynamics and impact dynamics. Energid's software includes the ability to automatically provide joint-level control simulation for complex robotic mechanisms having virtually any mobile robotic mechanism and automatically supports any number of links and any number of bifurcations in link connections. Additionally it works with virtually any type of end effector, including point positioning, orienting, frame, and planar constraints.

For this project, Actin is accessible and extendable by NRL through an open language, defined using the Extensible Markup Language (XML). It is implemented as an object-oriented toolkit that can be extended through subclassing. Energid will provide software with an interface that NRL can use to simulate scenarios and measure performance.