← Back to NASA Technology Projects
Completed TRL 3 (started at 2, targeting 3)
The proposed research effort will provide autonomous landers the capability to land on rugged terrain, having applications space applications for lunar/planetary landers. The use of adjustable landing leg lengths, stereo vision and artificial neural networks (ANNs) to map terrain at the landing site and determine a target landing vehicle state (target position, orientation, and landing leg length), and an ANN-driven fuel optimal controller to achieve the target landing state will be investigated through simulation and hardware implementation. Target landing zones on lunar missions have traditionally been restricted to large plains/plateaus due to restrictions on lander guidance precision and HDA capabilities. Even with the improved precision that came with investigations of optical navigation for HDA using ANNs, selected landing sites are limited to geometrically simple areas. These restrictions limit opportunities for scientific discovery on lunar missions by confining payload delivery to these specific regions. A landing system that is capable of successful delivery to virtually any terrain would vastly improve opportunities for exploration and discovery on other terrestrial bodies. A Stereo Vision-ANN based landing configuration determination system (LCDS) with ANN optimal controls can provide these successful deliveries, providing new opportunities for scientific discovery in lunar regions of geometrically complex terrain. The proposed LCDS includes an ANN trained to determine a target vehicle state based on a pair of images from a stereo vision system pointed at the terrain. This target vehicle state includes a target vehicle position, velocity, orientation, and landing leg length. This target state is then used by the ANN fuel-optimal controller to land the vehicle. The tools required to investigate the proposed research questions include (1) a tool to generate training and test data for the LCDS, i.e. data sets containing image pairs given a specified camera distance linked to a target vehicle state; (2) a tool to generate fuel optimal trajectories to train and test the ANN optimal controller, i.e. state-action pairs; (3) a tool to simulate the lander dynamics and implementation of the LCDS and optimal controller; and (4) a hardware implementation of a stereo vision system with LCDS and optimal control. Tool 1 will be based on the image generator presented in previous HDA works, and Tool 3 will be developed in MATLAB/Simulink. Tool 2 is already available as free MATLAB toolboxes (e.g. GPOPS-II, OpenOCL, etc.). Tool 4 will be developed in parallel with the other tools. One example of a low-cost (<$2,000) hardware implementation option is a 6 Degree of Freedom robotic arm and a sandy test bed. Simulation test-cases will be chosen such that a preliminary assessment of the system performance and robustness can be made, giving an initial answer to the research question. Tool 4 serves as a technology demonstration of the LCDS and ANN optimal controller, allowing for investigation of the research question in the context of a real application.
Target landing zones on lunar missions have traditionally been restricted to large plains/plateaus due to restrictions on lander guidance precision and HDA capabilities. The proposed research effort will provide autonomous landers the capability to land on rugged terrain, having applications space applications for lunar/planetary landers.
Listed on TechPort itself — the most direct way to ask about this specific project.
This is early/mid-stage (TRL 3) — the most realistic path in is NASA SBIR/STTR, which funds small businesses and research institutions to develop technology aligned with NASA's needs (equity-free, phased funding). Check whether a current SBIR/STTR solicitation topic overlaps with this project's technology area, or contact the project directly (above) to ask.
None of these are guaranteed paths for this specific project — TechPort itself doesn't have an "apply" button. Reaching out to the contact(s) above with a specific question is usually the fastest way to find out what's actually open.