Dobot 1200000404 Pocket Go Portable Learning

The Dobot Pocket Go is a portable data collection platform designed to capture high‑quality robot manipulation movements for AI training and automation development. It combines precision motion capture, high-resolution video, and an integrated software stack to enable efficient dataset generation for imitation learning workflows.

The system uses a GoPro Hero 12 camera with visual‑inertial SLAM (ORB‑SLAM3) to record operator movements with accurate 6D pose tracking, generating spatially annotated datasets without manual labelling.

Supplied with both a handheld gripper and an arm‑mounted end-effector gripper, recorded demonstrations can be transferred directly to compatible Dobot Nova and CRA series robot arms for autonomous execution.

The platform supports high‑throughput data collection, enabling up to 4,000 demonstration datasets per day, and includes a full software pipeline covering data capture, trajectory processing, model training, and deployment.

The Dobot Pocket Go is used across robotics and AI research environments for imitation learning and dataset generation, as well as in industrial automation development where it supports tasks such as pick‑and‑place and assembly.

It is also suited to commercial and retail automation applications, laboratory and academic settings for robotics development and testing, and product prototyping for autonomous manipulation systems, with broader use in general object handling scenarios across industrial, commercial, and domestic environments.

• Portable data collection platform for robot manipulation and AI training
• GoPro Hero 12 camera with 5.3K video and 6D pose tracking via ORB‑SLAM3
• Up to 4,000 demonstration datasets captured per day
• Includes handheld and robot-mounted gripper kits (up to 2kg payload)
• Compatible with Dobot Nova and CRA series robot arms (sold separately)
• Complete software stack for data collection, training, and deployment