LiDAR has become a go to people measurement technology for Digital Mortar.  Although LiDAR has numerous use cases including autonomous vehicles, mapping and robotics, in this review series we discuss the merits of different LiDAR sensor modes explicitly from a people (and vehicle) crowd measurement perspective.  We only review sensors that we have deployed on real world retail, airport, transit, stadium, casino, etc. settings.

 

Feel free to reach out to us at contact@digitalmortar.com with questions, feedback, to request a sensor for review or if you would like to discuss what hardware might be right for your measurement needs.

 

Livox Mid-360

 

Ratings Summary

Livox Mid-360	Score (out of 10)
Beam Pattern	5
Beam Density	5
Coverage	4
Value	10
Power	10
Support	2
OVERALL	7

 

Overview

The Livox Mid-360 is a dome style 360 mechanical LiDAR sensor intended for short range (40m is the published range but 8-15m is the practical range for people measurement).

 

Beam Pattern (5/10)

The Mid-360 has a 360 view and excellent 59-degree field of view.  This coverage is achieved using a variable beam pattern cycle (Livox refers to it as a ‘unique rotating mirror hybrid-solid technology’).  It’s effectively a tradeoff between less beam density for broader field of view.

The variable beam pattern might represent both the sensors greatest strength and weakness.  Without it the sensor would have a far more limited field of view and not be useful for people measurement.  With it, object identification (including people) is more challenging.  Luckily some perception software companies, like Outsight, have met that challenge and incorporated appropriate point cloud processing strategies to handle the variability.  Prior to leveraging this sensor, it’s important to check with your perception software to verify that they can correctly process variable beam pattern sensors.

While the 59-degree field of view is excellent, it does leave a blind spot directly underneath the sensor when mounted inverted.  The blind spot is much smaller than in typical puck sensors, but other true dome sensors, like the Robosense Bpearl or Ouster Dome, have no blind spot whatsoever.

The beam pattern makes this sensor well equipped for people measurement.  It’s perfect for ceiling/drop pole mounting (as long as the blind spot is factored in and can be covered by adjacent sensors) since the coverage is focused above the horizon of the sensor.

The small form factor combined with the beam pattern can deliver a lot of flexibility.  We are leveraging this sensor for under table applications, where one table/display is tracked by a single sensor mounted out of view under the display at waist height.   Under table or display embedded mounting strategies can also provide full journey coverage (consistent tracking of people across an entire space) by installing sensors under multiple tables/displays.

We like pairing up the sensor directly with a processor at/near the installation point.  That delivers a NO-CABLING solution that only requires power, since the processor can transmit data via Wi-Fi or cellular.  It provides a seamless way to deploy measurement without installation hassle and a low total cost of ownership when factoring in the low sensor price and lack of installation expenses. This is a perfect way to deliver sophisticated measurement in display or store-within-a-store concepts where measurement is desirable, but you don’t own the physical space or IT infrastructure.

 

Beam Density (5/10)

The Mid-360 touts delivering a ’40-line point cloud density’ but I suspect that line number is inflated by the (literal) use of mirrors.  The less impressive ‘200,000 points/s (first return)’ is the more telling and reliable density statistic.  Compared to the 5.2 million points per second of an Ouster OS0-128 that density sounds pathetic, but it can be serviceable as long as it’s used in context.

The sensor’s density limitation prevents it from being useful for long range people tracking but is plenty good within a 10-meter radius.  In full journey deployments, it would also be key to place adjacent sensors with the density in mind to get multiple vantage points on most/all objects moving inside the location.

Use of Mid-360s for large, open areas (where coverage density and radius are paramount) is not recommended but they are ideal for smaller, indoor environments with occlusions such as retail stores with shelving or custom displays.

 

Coverage (4/5)

Don’t bother looking at the 40m listed range on the sensor specs; in people tracking applications you’re not going to get near that number.  In our experience we are seeing between 8- and 15-meters radius of range around each sensor.  That range is toward the worst in class for LiDAR sensors but needs to be considered in the context of sensor cost and alternative technologies.

Compared to using stereoscopic cameras for people tracking, a Mid-360 has a similar per sensor cost but in a worst-case scenario will cover 2x the square footage of a best-case stereoscopic sensor (2,000 sq ft vs 1,000 sq ft).  From a comparative cost perspective, it would be possible to procure 5 to 10 Mid-360 for the cost of a single LiDAR sensor from another manufacturer with a better field of view.  Head-to-head the coverage of a Mid-360 would lose by a wide margin but if you put 5 or 10 together it is likely you get a better field of view for the same per-sensor cost (although don’t forget to factor in the differing installation and perception software costs).

 

Value (10/10)

These are the lowest cost sensors Digital Mortar uses for people measurement – by a fairly wide margin. Livox has broken new ground here in lidar price/performance for people-measurement. They are not in consideration for every, or even most, LiDAR based people measurement deployments but, in some instances, where you are measuring a focused area or interior areas with occlusions, they may deliver 100% equivalent performance to other sensors for 25% (or even less) of the cost. This is the first lidar system that can truly be competitive with camera in point measurement solutions. The Mid-360 is particularly attractive for field deployments where only one or two sensors are required, and no wiring is available.

 

 Power (9/10)

Little sensor, little power.  With only a 6.5W draw they don’t take much power to operate.  That means that they are compatible with every version of PoE — in comparison to the more powerful LiDAR sensors that require PoE+ if not PoE++.  If you go the route of using PoE for a multi sensor deployment, you likely won’t have to worry about exceeding your switches available power budget.

 

Support (2/10)

This is a new category to the LiDAR sensor review series that we created especially for Livox!   In our experience, most LiDAR manufacturers over support their products.  The emerging nature of the technology and the desire to make their sensors successful leads to phenomenal support from manufactures like Robosense, Ouster and Quanergy.

Livox is different and not in a good way.  They are elusive and prone to pass off to other departments (that may or may not have appropriate answers).

As an organization, Livox has all the appropriate support and communication channels, but they are too frequently ineffective.  Their focus is on building/pricing sensors at a commodity level which translates into the high value mark but, unfortunately, they also support them like a commodity.

Our intention here is not to bash Livox. We really like what they are doing in the market and the Mid-360 has opened up new opportunities for Lidar measurement that simply didn’t exist before. But if you’re thinking about these sensors, you need to know that support challenges are a risk.

 

A Note on Noise

Something to be aware of is that the Mid-360 creates more vibration/noise than other LiDAR sensors we have tested.  We suspect the cause of the vibration is the mechanism for creating the variable beam pattern but that’s just our guess.

The noise generated is variable based on the mount surface (and how much of the vibration gets translated to the surface).  Under no conditions is it ‘loud’ but the sensor is more audible than most in quiet settings.  In any type of environment where ambient noise is present (store, transit, airport, etc.) or the sensors will be away from people (i.e., ceiling mounted) the noise generated is negligible, but we might think twice about mounting this sensor under a desk/table in a quiet office, library or museum environment.