Teaching Collaborative Robots to Think and Learn Like Humans
I am working closely with the vision and firmware teams at Peer Robotics to enhance the Peer3000's intelligence and learning capabilities. I implemented a high-performance data pipeline that captures 30+ synchronized camera, sensor, and telemetry streams during operations. The data is processed in the cloud to train and fine-tune deep learning models (e.g., for object detection) to improve overall perception and decision-making.
Working across R&D facilities in Gurgaon (India) and Detroit (United States), as well as the manufacturing facility in Pune (India), I explore utilizing modern vision models, VLMs, and world models for hierarchical planning to enhance situational awareness, localization, and path planning. The goal is to increase the contextual awareness of the Peer3000, enhance proactiveness and improve autonomous decision-making in real-world manufacturing settings.
The Peer3000 is not just a tool that follows instructions. It is a coworker that understands its surroundings, adapts to new environments and learns just like a human colleague.
Watch more Peer3000 videos here.