Stretch - ROBOTS: Your Guide to the World of Robotics

Stretch is a compact, lightweight mobile robot equipped with a distinctive telescoping arm. It’s used as a research platform, and the latest version is also aimed at applied R&D labs and assistive-technology pilot projects exploring robot applications for homes and workplaces.
Hello Robot cofounders Aaron Edsinger and Charlie Kemp met at MIT, where they were students of famed roboticist Rodney Brooks.
Hello Robot introduced Stretch RE1, a compact, low-cost mobile manipulator designed for researchers, in July 2020. With offices in the San Francisco Bay Area and in Atlanta, Ga., Hello Robot was founded by ex-Googler Aaron Edsinger and Georgia Tech robotics professor Charlie Kemp. Previously, Edsinger was a cofounder of Meka Robotics and Redwood Robotics, which were acquired by Google in 2013, along with six other robotics companies, including Boston Dynamics. Edsinger stayed at Google as one of the directors of its robotics program until 2017, when he left to cofound Hello Robot with Kemp. Kemp heads Georgia Tech's Healthcare Robotics Lab, where he conducts research on mobile manipulators to assist older adults and people with disabilities. Hello Robot launched Stretch as a research platform but the company says it hopes to see the robot deployed in homes and workplaces so that it is helpful to people in society. Stretch 4 was introduced in May of 2026.
Specs for version 4. Lightweight design featuring a low-mass, contact-sensitive body. Safe to interact with people. Small footprint allows it to work in tight and cluttered environments. Head-mounted sensing system can see in all directions regardless of arm position. Ambidextrous arm can operate on either side of the robot. Supports advanced teleoperation and development of autonomous behaviors. Open source code. Extensible hardware.
• Head lidars: Dual 3D hemispherical Hesal J128 lidars 360° x 187° H x 189° FOV.
• Head wide FOV cameras: Dual global shutter 2.3MP RGB Cameras (Luxonis OAK-FFCAR0234).
• Head high-res camera: 12MP RGB Camera (Luxonis OAK-FFC IMX378).
• Wrist depth sensing: Stereo depth camera (Luxonis OAK-D SR) with 4 TOPs processing.
• Hazard detection: 6x Pixart laser line sensor curtains for cliff and hazard monitoring.
• Mobile base sensing: 9 DOF IMU for tilt detection and odometry.
Four Feetech smart servo actuators for the 3 DoF wrist and the 1 DoF gripper. Five custom closed-loop stepper motors for the 3 DoF omnidirectional mobile base, the 1 DoF lift, and the 1 DoF telescoping arm.
Arm with custom carbon fiber links. Head, shoulder, and mobile base have polyurethane covers.
Base: Steel and ABS plastics. Lift and arm: Aluminum. Head: Aluminum, polycarbonate, ABS plastics. Wrist: ABS plastics.
Primary compute: Intel NUC 15, Intel CoreTM Ultra 5 225H, 32.0 GiB RAM, 1.0 TB NVMe. Auxiliary compute: NVIDIA Jetson Orin NX, 16GB RAM, 128GB NVMe. Connectivity: WiFi 5.2 and Ethernet.
Ubuntu 24.04, ROS 2 Jazzy Jalisco. Code is open source and available on GitHub. Custom code written almost entirely in Python and firmware based on open source Arduino sketches. Open-weight AI models (including the Molmo 2 VLM, SAM 3 for segmentation and tracking, and RTMO for human pose estimation).
512Wh LiFePO4 battery with quick-swap capability, integrated charger and BMS, 8 hours of operation.