Stanford University research group (Tony Z. Zhao et al.) that originated the open-source ALOHA bimanual teleoperation platform and the ACT imitation-learning method. The hardware is commercialized as a kit by Trossen Robotics.
Mobile ALOHA is Stanford's low-cost whole-body bimanual teleoperation and imitation-learning reference design: an ALOHA-style leader-follower dual-arm rig (2 leader + 2 follower 6-DOF arms) mounted on a wheeled mobile base, letting a single operator puppeteer both arms and drive the base simultaneously to collect demonstrations for mobile manipulation tasks (cooking, dishwashing, taking an elevator). Co-training with the earlier stationary ALOHA dataset raises success rates by up to ~90%. Hardware BOM and ACT training code are open-source.
ALOHA (A Low-cost Open-source Hardware system for Bimanual Teleoperation) is a stationary tabletop leader-follower rig from Stanford (Tony Z. Zhao et al.), built for fine-grained imitation learning. Two ViperX 300 6DOF follower arms are puppeteered by two WidowX leader arms; a human operates the leaders and the followers mirror the motion while multi-view cameras record demonstrations. Paired with the ACT (Action Chunking with Transformers) policy, it became the most-cited low-cost bimanual data-collection platform from 2023 onward. The design is fully open-source (3D-print parts, assembly tutorial, training code) and is also sold as a kit by Trossen Robotics for roughly a $20k build budget.
Mobile ALOHA is Stanford's low-cost whole-body bimanual teleoperation and imitation-learning reference design: an ALOHA-style leader-follower dual-arm rig (2 leader + 2 follower 6-DOF arms) mounted on a wheeled mobile base, letting a single operator puppeteer both arms and drive the base simultaneously to collect demonstrations for mobile manipulation tasks (cooking, dishwashing, taking an elevator). Co-training with the earlier stationary ALOHA dataset raises success rates by up to ~90%. Hardware BOM and ACT training code are open-source.
ALOHA (A Low-cost Open-source Hardware system for Bimanual Teleoperation) is a stationary tabletop leader-follower rig from Stanford (Tony Z. Zhao et al.), built for fine-grained imitation learning. Two ViperX 300 6DOF follower arms are puppeteered by two WidowX leader arms; a human operates the leaders and the followers mirror the motion while multi-view cameras record demonstrations. Paired with the ACT (Action Chunking with Transformers) policy, it became the most-cited low-cost bimanual data-collection platform from 2023 onward. The design is fully open-source (3D-print parts, assembly tutorial, training code) and is also sold as a kit by Trossen Robotics for roughly a $20k build budget.
2 products