Deep Robotics Lite3 Venture

The Lite3 Venture adds the integration interfaces the Basic lacks: dual Ethernet ports and an external power supply with 24 V, 12 V, and 5 V outputs for hosting your own sensors and payload computing, in a slightly taller 17.5 inch (445 mm) body. The locomotion hardware is otherwise the same as the Lite3 Basic: it does not add an onboard AI computer, depth camera, or other perception sensors, although an NVIDIA Jetson Xavier NX perception host is available as an add-on through the onboard interfaces. Choose the Venture if you plan to integrate your own compute and sensors.

Only the base sensing. The Venture ships with the same sensing as the Basic: a built-in 9-axis IMU and joint encoders for balance, plus the wide-angle camera and ultrasonic radars of the base platform, which provide front and rear obstacle stop and visual following. It does not include a depth camera, lidar, or an onboard AI computer. You can add your own perception hardware through the dual Ethernet ports and the 24 V, 12 V, and 5 V external power outputs. Deep Robotics itself uses the Intel RealSense D435i depth camera and Livox or Leishen lidar modules on the Pro and LiDAR configurations.

The Venture does not include an onboard AI computer as standard, so you would add your own compute module, for example an NVIDIA Jetson, connected through the dual Ethernet ports and powered from the 24 V, 12 V, or 5 V external supply; Deep Robotics also offers a Jetson Xavier NX perception host as an add-on. The locomotion controller exposes a standard interface, so an external computer running your vision or policy code can publish velocity commands while the robot handles balance and gait. The open C++ and Python SDK supports this integration.

The Pro adds hardware the Venture does not have: an onboard NVIDIA Jetson Xavier NX perception computer, a factory-installed Intel RealSense D435i depth camera, forward obstacle avoidance, and USB 3.0 and HDMI interfaces alongside Ethernet and external power. The Venture is the Basic platform with dual Ethernet and multi-voltage power outputs for your own sensors and compute, and no added perception hardware. Choose the Venture to build your own perception stack at a lower cost; choose the Pro if you want onboard perception and compute out of the box.

You develop on a Linux laptop running Ubuntu 20.04 and communicate with the robot over Wi-Fi or Ethernet using the Deep Robotics SDK. The locomotion controller is exposed through the SDK and ROS, so your nodes send velocity and gait commands while the locomotion layer handles balance. ROS 1 (Noetic) and ROS 2 (Foxy) packages are published on Deep Robotics' GitHub.

Visual following is included: the Venture's baseline perception features are front and rear obstacle stop and visual following using the built-in wide-angle camera. Autonomous navigation is achievable but requires you to add perception hardware and compute, since the Venture does not include a depth camera, lidar, or an onboard AI computer. SLAM-based navigation requires a lidar and a navigation stack such as Nav2, which is how Deep Robotics' own LiDAR configuration works. Deep Robotics publishes perception development examples, but autonomous navigation is not a turn-key feature on the Venture.

Continuous walking runtime is approximately 1.5 to 2 hours per charge, the same as the standard Basic, since the Venture uses the same 28.8 V, 4.4 Ah hot-swappable battery. Charging from empty takes about 40 minutes to 1 hour. If you add your own compute and sensors, their power draw will reduce runtime accordingly. A second battery enables continuous operation through hot-swapping.

No. The Venture uses the same locomotion controller as the Basic, with the same 2.5 m/s top speed, 40 degree slope rating, and 5.9 inch (15 cm) step height, or 7.1 inches (18 cm) in AI Motion Mode. The mechanical platform, motors, and 12 degrees of freedom are identical between the two; rated payload is 9.9 lb (4.5 kg) and range up to 2.5 miles (4 km) on the Venture because of its added interfaces. The difference is the added connectivity, not the gait software.

Yes. Deep Robotics positions the Lite3 for academic research, reinforcement-learning experiments, and robotics education, and universities use it for legged-locomotion work. The open SDK and standard hardware make experiments reproducible by other labs. Cite Deep Robotics' platform documentation and include URDF version numbers in your methods sections to support reproducibility.

12-month manufacturer warranty on the electronics from the date of delivery, with the joints, legs, and battery covered for 6 months, reduced to 3 months once AI Motion Mode is unlocked. RoboticsSelect handles all warranty claims as the authorized dealer. The shell, accessories, normal wear on consumable parts, and damage from misuse, drops outside the recommended operating conditions, or unauthorized modification are not covered.