Deep Robotics Lite3 Pro

The Pro ships ready-to-use with three things the Basic lacks: an onboard NVIDIA Jetson Xavier NX perception computer, a factory-installed Intel RealSense D435i depth camera with forward obstacle avoidance, and USB 3.0, HDMI, Ethernet, and 24 V, 12 V, and 5 V external power interfaces. The chassis, motors, locomotion controller, and 28.8 V, 4.4 Ah battery are identical between Basic and Pro, so runtime is the same 1.5 to 2 hours. Pick Pro if you want autonomy and perception working out of the box.

The forward-facing Intel RealSense D435i depth camera feeds a forward obstacle-avoidance function that steers around obstacles in the robot's path, while the wide-angle camera and front and rear ultrasonic radars provide obstacle auto-stop in both directions. RGB-D point clouds are published on standard ROS topics. The perception host supports real-time Visual SLAM from the depth stream and CUDA-accelerated inference with PyTorch, TensorFlow, and models such as YOLOv8. Visual following and YOLOv8-based people tracking are provided as demo applications.

Not out of the box. Deep Robotics' turn-key Auto Mode for autonomous navigation is exclusive to the Lite3 LiDAR configuration. The Pro gives you the building blocks: Visual SLAM from the depth camera, forward obstacle avoidance, and the standard ROS navigation stack (move_base on ROS 1, Nav2 on ROS 2) on the Jetson, so you can map a space and send goal poses yourself, but you are integrating that pipeline rather than switching it on. For reliable building-scale autonomous navigation, choose the LiDAR configuration.

Maximum walking speed is 8.2 ft/s (2.5 m/s) on flat ground. Practical speed during autonomous navigation is well below that so obstacle avoidance has time to react. Deep Robotics does not publish a control range for the included controller; keep the robot within line of sight, or send commands over a network link from your own development host. Operating range per battery is approximately 2.1 miles (3.4 km) on flat terrain.

Yes. Stair climbing is rated for steps up to 5.9 inches (15 cm), or 7.1 inches (18 cm) in AI Motion Mode, and slopes up to 40 degrees are rated using the Grip gait. The robot uses its IMU and joint torques to detect step edges. For unfamiliar staircases, switch to the H-Step gait, which is intended for higher steps, and test at low speed with a spotter before running the full flight.

No. Deep Robotics does not publish an official IP rating for the Lite3 Pro. The manufacturer's manual advises against operating it in rain, snow, fog, or wet or humid conditions and warns that moisture intrusion can cause damage. Treat it as a dry-environment platform for indoor and controlled outdoor use: wipe away any incidental moisture promptly, avoid standing water and sustained precipitation, and store it dry.

The Pro publishes camera feeds, point clouds, joint states, and IMU data on standard ROS 1 or ROS 2 topics, and the motion host exposes a UDP interface. Inspection platforms that consume ROS topics can be wired in directly. Deep Robotics does not publish pre-built MQTT, REST, cloud, or historian connectors for the Lite3, so bridging to those systems is Python or C++ development on the Jetson.

Rated payload is 8.8 lb (4 kg) distributed, and that figure already accounts for the factory camera and Jetson package, so it is the budget available for additional sensors, small robotic arms, or task payloads. Mount points are M5 threaded inserts on the back deck. Off-center loads reduce stair-climbing capability so balance them as best you can.

Continuous walking runtime is approximately 1.5 to 2 hours per charge; running the camera and SLAM continuously sits toward the lower end. Charging from empty takes about 40 minutes to 1 hour. The battery is hot-swappable and a second battery is sold separately. For long shifts, plan on battery rotation through the charger.

Yes within the 8.8 lb (4 kg) payload limit. The Pro has Ethernet and USB 3.0 ports and 24 V, 12 V, and 5 V external power outputs that can power a small arm directly. The combined system requires you to coordinate the arm controller with the locomotion controller through ROS, since the two ship separately.

12-month manufacturer warranty on the electronics, including the control system, Jetson compute module, and factory-installed camera and radars, from the date of delivery. The joints, legs, and battery are covered for 6 months, reduced to 3 months once AI Motion Mode is unlocked. The shell, accessories, and normal wear on consumable parts are not covered. RoboticsSelect handles all warranty claims as the authorized dealer.