I work at a warehouse automation company and handle sensor selection. Last year we started upgrading the AGV perception system. We'd been using a 2D single-line LiDAR — it could only scan one horizontal plane, so pallet feet under shelves, stray objects on the floor, none of that showed up. After switching to 3D, things improved dramatically enough that I wanted to write down what we went through.

The LiDAR we went with is the M360 3D LiDAR from Tantu Smart Mobility, a Chinese manufacturer. A few things stood out during eval, which I'll get into.

Installation and Calibration

First, the physical install. The M360 measures 78×78×81mm and weighs 408g — not bad. We mounted it underneath the front of the AGV, tilted downward at roughly 15°. Why the tilt? The vertical FOV is 70°, spanning from -10° to +60°. At that angle, both the ground and the shelves ahead are covered. We'd previously used a LiDAR with a 59° vertical FOV at the same angle, and the ground coverage was noticeably worse.

After mounting, the first order of business was coordinate calibration. The M360 saved us some real work here: it has a built-in 6-axis IMU and supports PTP v2 time synchronization. Point cloud data and IMU timestamps sync at nanosecond level. With other LiDARs we've used, you'd need an external IMU module and sort out timestamp alignment yourself — not terrible, but fiddly and easy to get wrong. The M360 skips that entirely. Data consistency during coordinate transforms was solid right from the start.

Point Cloud Processing

With calibration done, we moved to point cloud processing. The M360 outputs at 200kHz using a non-repetitive scanning approach with a rotating mirror. Unlike conventional multi-beam mechanical LiDARs, each revolution covers different vertical angles. In our warehouse, AGVs typically move at 0.5–1 m/s through rack aisles. Standing still for a second or two fills in the point cloud noticeably. For obstacle avoidance, this is actually useful — by the time the AGV slows for a turn, the point density is already sufficient.

Avoidance Setup in Practice

We use DWA (Dynamic Window Approach) for obstacle avoidance. The pipeline is straightforward: downsample the M360 point cloud, inflate obstacles within a 3-meter radius around the robot, then feed everything into the DWA scoring function. A few details worth mentioning.

First: blind zone. The M360's near blind zone is only 5cm. That's smaller than most 3D LiDARs on the market. On a warehouse floor, you regularly encounter bolts, cables, and pallet feet — things in the 2–5cm height range. With our previous LiDAR (10cm+ blind zone), these just didn't register. The AGV drove over them more times than I'd like to admit. After switching to the M360, ground obstacles beyond 5cm are reliably detected.

Second: rain, fog, and dust. Part of our warehouse sits near the loading dock. During busy hours, the dock doors stay open and rain drifts in. The old LiDAR would produce a noticeable spike in noise — the point cloud would be full of scattered returns. The M360 supports a dual-return mode that can separately receive forward and backward echoes, seeing through raindrops to the objects behind them. After switching to dual return, rain-day noise dropped significantly. There's also a built-in rain/fog detection function that automatically adjusts point cloud output based on conditions.

Third: IP protection. This one's a bit embarrassing to admit. Our previous LiDAR didn't have a rated IP class, and we didn't think much of it at the time. That changed when an AGV rolled through an area a cleaner had just mopped. Water splashed onto the chassis, and the LiDAR threw an error within minutes. The M360 is IP67 — fully dust-proof and rated for temporary submersion. Since installing it, we've stopped worrying about damp or dusty warehouse zones.

Power and Runtime

One more thing on power consumption. The M360 draws under 4.5W. Our AGVs run on 48V lithium packs, and the previous LiDAR pulled around 8W. A 3.5W difference might not sound like much, but AGVs typically run 8+ hours per shift, and with two shifts per day, the power difference matters for battery life and thermal design. Lower power means smaller heatsinks and a more compact overall build.

After Over Six Months

We've had the M360 running in the warehouse for over six months now. No major obstacle avoidance issues to report. The two problems that used to come up constantly — missed small ground obstacles and rain-day noise spikes — are basically gone. The built-in IMU meant we didn't need to deal with external modules, and the IP67 rating is one less thing to worry about in a warehouse.

If you're in the middle of a sensor eval for AGVs, the M360's 5cm near blind zone, dual-return rain handling, IP67, and built-in IMU with PTP sync are genuinely useful in warehouse settings. Not marketing fluff — stuff we actually rely on day to day.

*Data source: M360 product manual (Tantu Smart Mobility) and actual deployment experience. Performance may vary — refer to the latest official specs.*