Software and Hardware Experiments

Software

  1. SCP Toolbox - Julia-based library implementing state-of-the-art optimization-based methods for trajectory generation. Developed in collaboration with Danylo Malyuta, Taylor P. Reynolds, Michael Szmuk, Thomas Lew, Marco Pavone, and Behçet Açikmeşe, this library implements sequential convex programming, offering efficient and reliable solutions to several complex real-world control problems, such as the rocket landing problem. This open-source and user-friendly toolbox is accessible at the following GitHub repository.

  2. PBDS - Julia-based library implementing Pullback Bundle Dynamical Systems, a differential geometric paradigm for real-time policy generation. Developed in collaboration with Andrew Bylard and Marco Pavone at Stanford University, this library enables computing composed policies in the range 300-500 Hz, for complex, high-degree-of-freedom robotic systems operating in cluttered environments. The whole open-source library can be found at the following GitHub repository.

  3. SOCP - C++-based industrial software implementing indirect shooting methods for real-time trajectory generation of endo-atmospheric, thruster-based systems. I developed this software in collaboration with Bruno Hérissé at ONERA. It can compute optimal solutions to complex non-convex endo-atmospheric rendezvous problems in few milliseconds, on onboard computers of few kilobytes of memory. Although owned by ONERA, an open-source, beta-version of this software can be found at the following GitHub repository.

Hardware experiments

  1. Hardware experiments aboard the International Space Station - Developed in collaboration with Abhishek Cauligi and Marco Pavone at Stanford University, through these experiments we successfully validated some of our algorithms for grasping maneuvers on real NASA’s space robots Astrobee. You can witness a portion of these validations in this video.
  2. Hardware experiments at the Stanford Space Robotics facility - Developed in collaboration with Thomas Lew and Marco Pavone at Stanford University, through these experiments we stress-tested some of our safe-against-uncertainty optimal control algorithms. Specifically, as demonstrated in this video, we made realistic replica of space robots safely navigate simulated two-dimensional micro-gravity uncertain environments, outperforming popular state-of-the-art methods.