Shield AI Hivemind simulator

According to Shield AI’s description of the video, Hivemind uses planning, mapping and state estimation to fly dynamic maneuvers, and reinforcement learning to discover tactics. It is designed to run on the aircraft in environments where GPS and communications are degraded.

The flight shown in the video runs in the simulator I developed.

My work at Shield AI since 2018 also covers:

• Synthetic datasets for training machine learning models, and virtual environments for training robots
• Scalable robot-swarm simulation with Unreal Engine, PX4 and ROS
• Autonomous agents for tactical mission planning, scenario generation and concept-of-operations environments

In 2018 Shield AI interviewed me about the overlap between games and AI.