I'm a mechatronics engineer specializing in control systems, robotics, and the application of physics principles to solve complex technological challenges. Currently, I'm the Head of Technology at Summon, pioneering advancements in humanoid robotics.
Building the future, one system at a time
Hello! I'm Sanket, an engineer interested in the intersection of technology, mathematics and physics. What began as curiosity about how things work has grown into a drive to build systems that can move and think in the real world, blending control theory, physics, reinforcement learning and hardware design.
I've worked on non-GPS navigation for high-altitude UAVs and on both electronics and software for bipedal robots. I enjoy open-source collaboration to tackle challenges from space exploration to terrestrial automation.
I also dabble in high-energy particle and quantum physics; after all, who doesn't like questioning reality before breakfast?
Here are a few technologies I've been working with recently:
My professional journey
Leading and developing electronics and software stack for humanoid for cleaning industry. Working on Isaac Sim, Isaac Lab, VLA models, RL, PINN (learning based controller), Classical & Modern Control systems, electrical circuit design, power management for bipedal humanoid. Developing real-time AI assisted long range teleoperation system and working on genetic algorithms, learning by cheating, H-GAIL, Actor-Critic, VLAs, ACT for bipedal legged control and bimanual learning for mobile manipulation.
Developed the navigation and controls stack for the Underwater AUV under IdEX project. Developed PINN based MPC, behaviour trees, and various controllers and state estimators. Leading the computer vision, controls and mathematical modelling part of the High Altitude Aerostat for Indian Defence under IdEX project.
Lead the software development of collision-free swarm algorithms for drone-show stack. Worked on modifying and developing Ardupilot codebase as per company requirements. Worked on perception, trajectory planning, control designs and firmware development. Worked on long range communication and antenna modelling for one-to-multi communication.
Developed High Altitude Non-GPS Position Estimation algorithm for UAVs. Designed and implemented transformer and optimisation based localization and mapping algorithm. Developed custom algorithms using computer vision and deep learning. Technologies: C++, Python, OpenCV, PyTorch, Lua and Embedded software development.
Developed a low cost precision landing solution. Working on a monocular camera and offboard computer based approach. Deployed and used image processing, pymavlink and autonomous capability on UAV.
Developing a terrain following algorithm for Agriculture UAVs. Developed simulated worlds and multiple terrain models. Used and learnt about Advanced Control Systems and advanced robotics. Deployed and used image processing, control systems and autonomous capability on UAV.
Updated ROS integration for Non-GPS navigation and autonomous path-planning. Developed a LiDAR equipped quadrotor with Ardupilot FCU. Updated the base code for the Cartographer SLAM tuning parameters and control algorithms. Integrated the offboard object avoidance with ArduPilot Auto mode.
Featured projects and contributions
Contributions to the Guidance, Navigation, and Control (GNC) and Attitude & Orbit Control Systems (AOCS) for an open-source space station operating system. My work involves designing and implementing the behavior tree architecture for autonomous operations and GNC & AOCS physics backend and SSOS SITL.
A comprehensive software and hardware stack for a bipedal humanoid for Summon. The product involves real-time AI-assisted long-range teleoperation, advanced control systems (RL, PINN, VLAs), and bimanual mobile manipulation, tested & simulated in Isaac Sim and deployed on hardware.
A novel GSoC project for Ardupilot to enable UAV position estimation at high altitudes without GPS. It leverages a downward-facing camera, fusing a transformer-based Visual Positioning System (VPS) with Visual-Inertial Odometry (VIO) for precise localization against satellite imagery.
Designed and implemented ArduPilot-based synchronized drone swarm control and long-range wireless comms for drone light shows.
National finalist entry for the ISRO CANSAT competition. Developed a satellite with active 2-axis attitude stabilization using reaction wheels and LQR controllers.
Built a monocular vision-based landing system for UAVs using PyMavlink, grid map, and OpenCV for high-accuracy landings.
An old age nursing ROS based humanoid robot capable of autonomously mapping and navigating around the house and providing medicine and water to elderly people. Developed whole perception software and 3D printed the face.
A 4-legged autonomous robot capable of giving medicines, walking and having offboard AI to interact with users. Kinematic and control model for motion written from scratch on indigenous microcontroller, Vega processor.
Notable accomplishments and competitions
National Finalist for developing a CANSAT with active attitude stabilisation using reaction wheels and LQR controllers.
India Regional Finalist for developing a blockchain based drone aggregator platform.
Led a team of 6 members to the National Finals organized by Ministry of Education (MoE), Government of India.
National Level Runner-up in IIT Bombay organized competition for a custom flight controller and program developed for a ROS based delivery drone.
Winner at Chandigarh University organized competition for Sahaya humanoid project.
1st Rank winner in multiple international hackathons: Quellie Hacks, StemWarriorHacks, RoboHacks and HackKu in hardware and overall categories.
Academic background
Chandigarh University
CGPA: 8.12/10
Activities: IEEE RAS Treasurer
Participated and won multiple recognized national and international robotics competitions
Giving back to the community
Actively contributing and mentoring contributors in the Ardupilot community, focusing on projects like, non-GPS absolute LiDAR positioning and the implementation of Software-in-the-Loop (SITL) with Reinforcement Learning for advanced autonomous systems.
Sharing knowledge with the community
I'm always seeking new challenges and opportunities to collaborate on ambitious projects. Whether you have a question, a proposal, or just want to say hello, my inbox is always open.
Say Hello