This project, developing a platform for research and training to serve as an instrument for evaluating robotic subsystem and supersystem performance, accelerates and enhances the evolution of autonomous vehicle subsystems and component parts by providing a baseline instrument to measure and assess the performance and capabilities of the various portions of a given autonomous vehicle design. This Autonomous Robotic Vehicle Instrument (ARVIN) is designed to allow simultaneous, parallel operation of multiple instruments, actuators, and software/hardware subsystems. Due to the common environment in which the items are operating, the ARVIN yields a precise, robust metric of comparative performance and capability of the item in question. Inherently rapidly reconfigurable by design of its network-centric architecture, the ARVIN enables quick, easy substitutions and augmentations of sensors and actuators on the evaluation platform; it is not a set of identical robot structures that join and reform to move or climb. The ARWIN tests, simulates, and validates subsystems and components in three tightly coupled hierarchical areas acknowledge as critical to advances in robotics vehicles: -Guidance, Navigation, and Control (GNC), -Sensor and Actuators Suites, and -Software and Network Architectures. The Autonomous Robotic Vehicle Instrument consists of an Argo Conquest 6x6 Off-road/Amphibious vehicle, heavily modified, and a Hardware-in-the-Loop (HIL) simulator for the vehicle. The vehicle will be fitted with ultrasonic sonar and SICK LiDAR for collision avoidance, fiber optic gyros, IMU's, and NavCom STARFIRE differential GPS receivers for guidance, a Pentium class PC-104 stack with high speed communication networks for control, and a Videre MEGA-STC-VAR stereoscopic high speed color camera for vision capture. In general, due to the incredibly high cost (in time and money) of constructing an autonomous vehicle of any kind, designs are rarely modified unless the mission cannot be completed by the original vehicle design. Very rarely are the initial trade-off studies revisited and explored to see what might have been done differently. Thus, the ARVIN should have profound impact on the kind of trade-off studies that most autonomous vehicle design programs have to-date been missing.
Broader Impact: ARVIN allows students and researchers to gain experience on an actual, physical, autonomous rover. This kind of practical training in systems integration, sensor fusion, software architecture for real-time systems, and actual control systems implementations has much to offer. The Naval Postgraduate School and the Intelligent Robotics Group (IRG) at NASA-Ames has indicated interest and enthusiasm in collaborating and experimenting with the modular chassis that will result of the ARVIN. Contributing to further understand trade-offs for mobile robotics, the platforms will also be used for outreach. The ARVIN infrastructure should become a cost-effective general robotics toll that may be easily replicated at other institutions.
