A Recon gurable Testbed Environment for Spacecraft Autonomy

A key goal of NASA's New Millennium Program (NMP) is the development of technology for increasing spacecraft on-board autonomy. Achievement of this objective requires the development of a new class of ground-based autonomy testbeds that can enable the low-cost and rapid design, test and integration of the spacecraft autonomy ight software. This paper describes the development of an Autonomy Testbed Environment (ATBE) for the NMP Deep Space I comet/asteroid rendezvous mission. This simulation testbed has been designed to enable rapid design of ight modules, early identi cation of performance and design problems, resolution of integration issues, and thorough ground testing for reducing mission-risk. ATBE's simulation requirements span a wide range of engineering platforms, functional and delity models, failure modes, test scenarios, and durations. The ight software modules under development include attitude control subsystem, remote agent, autonomous navigation, and ight systems control. Conventionally, such testbed functionality has been met by the expensive and time-consuming development of multiple specialized testbeds. In contrast, the ATBE testbed has been designed to be recon gurable for multiple user development and test needs. The ATBE software will also be integrated with the support equipment for hardware-in-the-loop tests and system level integration. The ATBE spacecraft simulator includes a high delity real-time dynamics simulation package integrated with simulation models for several of the hardware devices and interfaces on the spacecraft. The testbed incorporates existing in-house and third-party software, integrated within an object-oriented architecture. This design enables easier maintainability and usability, and perhaps most signi cantly this exible design is geared to handle continual evolution in model requirements, functionality and delity. The simulation interfaces are highly con gurable to allow swapping in and out of hardware as needed. The testbed has been instrumented from the start to provide a high degree of visibility into the simulation status with capabilities to peek/poke/checkpoint/resume model states, and includes some graphical user interfaces as well.