Best LiDAR for a Unitree Go2 EDU: Mid-360 and alternatives compared

For most Unitree Go2 EDU projects the best add-on LiDAR is the Livox Mid-360: it weighs about 265 g, draws about 6.5 W, sees 360° horizontally and −7° to 52° vertically, and has mature ROS 2 drivers and SLAM support. Pick the Mid-360S if you need more range, a Hesai JT16 if every gram matters, and a wide-FOV sensor such as the RoboSense Airy if near-field coverage is your priority. Below is how we choose, with weights and power taken from manufacturer datasheets.

Do you need an extra LiDAR at all?

The Go2 EDU ships with a Unitree L1 4D LiDAR in the head. It has a very wide 360° × 90° field of view and is good for obstacle detection and stair perception, but its rated range is about 20 m and its point density is modest. For indoor demos and short corridors it can be enough. For mapping factories, warehouses, tunnels or outdoor sites, for repeatable autonomous routes, or for LiDAR-inertial SLAM such as FAST-LIO or LIO-SAM, a second LiDAR on the back gives longer range, a clear view above the robot body and much better localisation in large spaces.

Comparison: LiDARs that fit a Go2 EDU

All values are from the manufacturers' published specifications as collected in our component catalog. Ranges are the rated maximum; usable range on dark or wet surfaces is shorter.

How to choose

SLAM and navigation (most common): Livox Mid-360. The non-repetitive scan pattern gives good coverage even when the robot stands still, the built-in IMU helps LiDAR-inertial odometry, and the open-source tooling around it (livox_ros_driver2, FAST-LIO, LIO-SAM forks, Point-LIO) is the most widely used on quadrupeds today. At 265 g and 6.5 W it costs very little of the Go2's payload and battery.

Larger outdoor sites: the Mid-360S keeps the same mounting and wiring as the Mid-360 with a longer rated range. The Ouster OS1 reaches much further but draws about 20 W, which noticeably shortens walking time on a Go2.

Lowest weight: Hesai JT16 at about 200 g and 4.3 W. Its vertical FOV is 0° to 40°, so it sees mostly upward; tilt the mount forward or rely on the head L1 for the ground in front.

Near-field coverage around the robot: RoboSense Airy or Ouster OS0, both with about 90° vertical FOV. The Airy is the better fit for a Go2 power budget.

Avoid on Go2 unless you have a reason: classic 16-channel spinning sensors such as the VLP-16. They work, but at around 830 g and with a tall cylinder they raise the centre of mass and are easier to damage in a fall.

Mounting, power and data

Placement: mount the LiDAR on the back plate, slightly forward of centre and high enough that the robot's own head and legs do not fill the lower part of the scan. Keep a clear 360° view around the window; antennas, cameras and handles next to it create blind sectors. A rigid mount matters for SLAM: any wobble shows up as drift.

Power: the Go2 EDU expansion port supplies battery voltage. Most of the sensors above accept a wide input range, but we still use a DC-DC converter and a small distribution board so the LiDAR, compute and other payload devices share one protected, fused supply.

Data: these LiDARs use Ethernet. On a Go2 EDU you either connect to the robot's internal network or put a small switch in the payload so the LiDAR, an onboard Jetson and any IP camera share one link. Use static IPs and, for Livox, set the host IP in the driver's JSON config.

Time sync and extrinsics: for good maps, synchronise the LiDAR and computer clocks (PTP where supported) and calibrate the transform between the back LiDAR, the IMU and the robot base. Publish it as a static transform in ROS 2 so navigation and mapping use the same frame tree.

Weight budget

A Mid-360 with bracket, cable and a small power board typically adds well under 1 kg. That leaves most of the Go2 EDU's payload capacity for compute (for example a Jetson Orin NX module), a camera or thermal gimbal, and a 4G/5G module. Keep heavy parts low and centred and leave margin instead of loading the robot to its rated maximum, so gait stability and run time stay acceptable.

Next steps

You can try LiDAR, camera, compute and communication combinations on a 3D Go2 model and see total weight and peak power in our online payload configurator, then send the configuration to us for a quote. See the Unitree Go2 EDU platform page and the Go2-W page for platform details. YUANZHI ROBOTICS designs and builds complete robot payloads, does quadruped development and integrates autonomous navigation and mapping on customer robots, including on an onboard computer you already have.

FAQ

What is the best LiDAR for a Unitree Go2 EDU?

For SLAM and autonomous navigation, the Livox Mid-360 is the most common and practical choice: about 265 g, about 6.5 W, 360° horizontal and −7° to 52° vertical field of view, with a built-in IMU and well-supported ROS 2 drivers.

Is the built-in Unitree L1 LiDAR enough for mapping?

For small indoor areas and obstacle avoidance it can be. Its rated range is about 20 m, so for factories, tunnels, outdoor sites or repeatable autonomous routes a second LiDAR on the back gives much better maps and localisation.

Can I run a Livox Mid-360 on my existing Jetson Orin NX on the Go2?

Yes. The Mid-360 connects over Ethernet and runs with livox_ros_driver2 on Ubuntu with ROS 2, so it can be integrated with a Jetson Orin NX you already have on the robot.

How do you power a LiDAR on a Go2 EDU?

From the Go2 battery through the expansion power port, normally through a DC-DC converter and a small fused distribution board shared with the other payload devices.

Who can integrate a LiDAR payload on a Go2 EDU?

YUANZHI ROBOTICS designs the mount, power and network wiring and sets up ROS 2 drivers, SLAM and navigation on Go2 EDU robots. You can build a configuration in the online configurator and request a quote.

Configure your payload and get a quote