We've been building autonomous forklifts for over three years. We started with 2D single-line LiDAR for navigation and gradually moved to full 3D perception. Last year we began testing the Tantu M360 alongside a Livox MID-360 as a reference. After six months in operation, here are our findings.

Autonomous forklifts are not your typical AGVs

Although they're both categorized as "unmanned material handling," autonomous forklifts operate very differently from warehouse AGVs:

These differences mean forklifts have distinct LiDAR requirements compared to standard AGVs.

Why we moved from 2D to 3D

We previously relied on 2D single-line LiDAR for navigation. It handled planar obstacle avoidance fine, but had one critical blind spot: it only scans a single horizontal plane. Anything above or below that plane is invisible.

One incident drove the point home. A forklift was reversing in an aisle, and the 2D LiDAR confirmed the path behind was clear. But a carton had been left protruding from the top shelf, just above the scanning plane. When the forklift tilted backward during mast movement, it knocked the carton down. Fortunately, no one was underneath.

After that, we committed to 3D. A 3D LiDAR covers vertical space — protruding items on shelves, hanging cables, and unevenly stacked pallets all become visible.

M360 vs MID-360 on forklifts

Both units were mounted on the forklift roof at 2.2m height, tilted downward at 15°. Each ran for two weeks in an A/B comparison.

Near blind zone

Forklifts typically maintain a 20–30cm gap from racking. Before the forks enter a pallet slot, the engagement zone needs to be obstacle-free. Typical obstacles here: deformed pallet crossbeams, stray packaging straps, debris left by a previous forklift.

The M360's 5cm near blind zone is a clear advantage. In our tests, for obstacles under 10cm near the fork area, the M360 detected roughly 95% versus the MID-360's 80%. The gap comes almost entirely from the 5–10cm range — the MID-360's 10cm blind zone swallows that entire band.

For autonomous forklifts that need 2–3cm engagement precision, a LiDAR that can't see anything between 5 and 10cm is a safety concern. Five centimeters is a comfortable lower bound.

Power and input voltage

Our forklifts use 48V or 80V battery packs with total vehicle power draw in the 3–5kW range. A few watts difference between LiDARs doesn't affect runtime.

But input voltage design matters. Our forklifts step down from 48V main batteries to 24V via DC-DC converters for sensor power. The M360's 12–32V input range provides generous headroom. The MID-360's 9–27V also works at 24V, but with minimal margin. Under heavy-load hard braking, voltage spikes can trigger undervoltage protection.

We learned this the hard way. The MID-360-equipped forklift experienced two LiDAR restarts during heavy-load deceleration. We had to add a voltage regulator module to fix it. The M360 unit never had this problem.

IP protection

Both are rated IP67. Forklifts deal with plenty of dust and occasionally pass through wet areas. IP67 is sufficient. Worth noting: the MID-360's IP67 rating was added in a recent hardware revision. Older batches may not carry it. Check the manufacturing date when ordering.

Non-repetitive scanning + low-speed accumulation

Forklifts move slowly in racking aisles and sometimes stop entirely for pallet alignment. Both LiDARs support non-repetitive scanning, and angular resolution improves during stationary periods.

In practice: after 3 seconds of stillness during pallet engagement, rack beam edges and pallet pocket contours become clearly visible in the point cloud. This directly helps docking accuracy.

The M360's 70° vertical FOV covers 11° more than the MID-360's 59°. At our 2.2m mounting height with 15° downward tilt, the M360 can sense down to 0.5m from the ground, while the MID-360 reaches roughly 0.8m. For close-to-ground perception, the M360 gives more complete coverage.

Built-in IMU

Both have built-in 6-axis IMUs with PTP v2 time sync. For forklifts, IMU is mainly used as a positioning supplement — GPS signal is unreliable indoors, so we rely on LiDAR + IMU for dead reckoning. Both are adequate. No meaningful difference in practice.

Dual return (M360-D)

Most of our forklift operation is indoors, so dual return rarely comes into play. But we have one scenario where it helps: on rainy days, forklifts carry goods from the indoor warehouse to trucks parked outside, passing through a semi-covered loading area with standing water.

In single-return mode, the point cloud over wet areas turns noisy and unreliable. With dual return enabled, the forward echo off the water surface gets filtered, and the return echo passes through to capture the actual ground or obstacles beneath. Perception quality over wet surfaces improves noticeably.

It's not an everyday feature, but it did prevent an incident where a forklift misjudged ground height over standing water and scraped its forks on the floor.

Our deployment

Our current fleet of 12 autonomous forklifts all use the M360 (four of them are the M360-D dual-return variant). The setup:

The MID-360 test unit stabilized after we added the voltage regulator. We ultimately chose the M360 over the MID-360 for three reasons: the 5cm blind zone matters for fork engagement safety, the 12–32V input range eliminates the need for extra regulation, and dual return helps in semi-covered outdoor areas.

Selection advice

For autonomous forklift LiDAR selection, these specs are worth prioritizing:

Both the M360 and MID-360 can meet the basic requirements of autonomous forklifts. The M360 has an edge in blind zone and power input range. The MID-360 is lighter and more compact. The right choice depends on your forklift's precision requirements and electrical design.

*Based on our own field testing. MID-360 specs sourced from Livox's official website; M360 specs from Tantu Smart Mobility. Results may vary by vehicle type and operating environment. Evaluate based on your own requirements.*