Low-Cost, Fault-Tolerant Controls for Unmanned Aircraft

Autonomous unmanned systems require provisions for fault detection and handling. Multiply-redundant schemes typically used in aerospace applications are prohibitively expensive for general applications, especially where low cost or small size are critical. A new strategy, proposed herein, offers the potential for low-cost fault-tolerant control of paramter-dependent systems that will be useful in a range of commercial and military applications. A "monitoring observer", or failure-detection filter, would run a real-time simulation of the controlled system in parallel with the actual system, compare the outputs of the two systems, and interpret discrepancies to detect system failures. The system would then be reconfigured to operate on a reduced sensor or actuator set. Both the failure detection and control reconfiguring take into account the parameter-dependent nature of the system. The core technology for this system--a method for tranforming linear-parameter-dependent systems such that they can be treated as linear-time-invariant for control design--has been demonstrated at MIT on an autonomous unicycle robot by the Principal Investigator. During Phase I, we will expand this work and develop a parameter-dependent failure-detection filter and control reconfiguration strategy suitable for use aboard unmanned aerial vehicles (UAVs).