Abstract
Powered descent vehicles (PDVs), such as the Mars Science Laboratory (MSL) and the Apollo Lunar Module, play an integral role in safely landing both robotic vehicles and human crews. The landing accuracy of PDVs has increased throughout the history of PDV development, with the current generation, MSL PDV, landing its payload within several kilometers of the intended target. However, future missions call for pinpoint landing, which requires PDVs to land within tens of meters of the target. The NASA Design Reference Architecture 5.0 for human missions to Mars calls for multiple assets to be deployed to the Martian surface prior to crew landings. These assets will need to be within tens of meters of each other to be reachable by crews. Additionally, there is a need for adaptable PDV guidance and control systems that are reconfigurable in-flight. A guidance and control strategy of this nature would enable a PDV to land safely in the event of component failure or performance degradation without the loss of crew or assets.
The ultimate goal of the research is to develop an autonomous guidance and control strategy that permits pinpoint landings in uncertain and dynamic environments, and is also robust to component failures. Towards meeting this ultimate goal, this work sets two objectives. First, a powered descent vehicle must be enabled to adapt in real-time to failures and degradations in its performance. An adaptive control allocation method is implemented that utilizes parameter identification techniques and inertial measurement unit data to identify if an engine has failed and what kind of failure it has experienced. Second, this research lays the groundwork for enabling a guidance routine to perform trajectory path re-evaluation and re-planning onboard in real-time. A guidance software is built that discretizes the trajectory design space, and evaluates each candidate trajectory using an internal six degree-of-freedom simulation. Once each trajectory has been simulated, constraints are verified and each trajectory is scored to determine a champion trajectory that is then passed to the control system to follow. This work leverages the high performance computing capabilities of multi-core central processing units by applying Open Multi-Processing directive-based parallelism techniques to parallelize the guidance software. This guidance software is then used to safely target and land a human scaled PDV (defined by the human Mars entry, descent, and landing architecture study) in several scenarios. These include avoiding keep-out-zones that the PDV cannot fly through, and diverting to secondary landing targets.
