Using Time of Flight Depth Sensor to Monitor Motion

Nowadays, monitoring of human motion can help to restrain a large group of people gathering at a place. As well as, monitoring of the robot’s movement can help to prevent collisions from happening.

Author: Ho Xiu Ting
Date: 27th May 2021

 

As clusters are growly rapidly throughout the community worldwide, it is important to limit the number of people entering a confined area. This is to maintain safe distancing from one another with the help of a sensor. For example, by mounting a sensor near the doorway, it starts to count the flow of the traffic and collects it to the database. With this information, the management of the mall can restrict the number of people entering. The case study shown below it will further explain the motion of humans.

 

What is Time of Flight?

Compared to other sensors that use ultrasound waves to compute the distance from object A to B, Terabee Evo 64px emits light to calculate the distance as it provides faster reading, greater range, and accuracy. When the visible light is emitted, the sensor will start to compute the time for the wave to reflect back to the receiver. Calculating distance will not an issue with the known speed of light and the time traveling to and fro using the formula below.

Distance = (Speed of Light * Time Travelled) / 2

Terabee Evo 64px is solely for indoor use only, as the sensor is sensitive to ambient light. With ambient light, the sensor would not be able to differentiate which protons were emitted by the sensor. Thus, resulting in having a technical issue with the data collection. This sensor can work in low light and complete darkness which makes it an ideal application for indoor usage. Evo 64px is eye-safe in all conditions, unlike laser sensors.

Figure 1: Matrix of Terabee Evo 64px

 

Privacy

Unlike the CCTV camera, the ToF technology that Terabee contains does not intrude into personal information. It only collects non-intrusive depth image data for sensing the distance/ counting the number of people. Hence, most of the ToF sensors are General Data Protection Regulation (GDPR) compliant by design and protect the privacy of all customers and staff. The gif below had proven that no image other than matrix is being taken from the camera sensor.

 

Terabee Evo 64px

As long as the sensor is placed at an angle of 15° Field of View, it can capture the motion of people passing through it. In this experiment, Terabee Evo 64px is positioned at the ceiling and at the side of the door to compare the difference in monitoring motion in a different position. Since the placement is different, a slight adjustment has to be made in the algorithm of the sensor.

    
(a) Sensor on the ceiling              (b) Sensor at the side
Figure 2: Positions of Terabee Evo 64px

Results have proved that when the sensor is placed on the side of the doorway, the tolerance level of detecting the motion will be smaller as compared to when the sensor is positioned on the ceiling. This will cause greater interference when adjusting the algorithm. Some of the information is not transmitted as the sensor at the side could not sense the difference in the values. Therefore, placing the sensor on the ceiling will be a better alternative.

When the experiment was conducted, data was being transmitted to both Excel and HTML files for data analysis and pre IoT purposes respectively. However, this is in the initial design phase whereby the data and the visual are still incomplete.

 

Excel Result

Random variables (Incoming or Exit) are not collected constantly if no one passes through this sensor. Thus, it is practically impossible to compute the mean and variance for all the time periods using time series analysis, unless it is a stationary process where time is constant. By doing a stochastic process in time series analysis, the management is able to conclude which is the peak period. This allows them to inform the public which timings to avoid the crowd if they must be in the mall. The number of visitors that enter and exit the mall is shown in the stack column chart below.

Table 1: Recording the number of people in each period

Count of Date	Column Labels
Row Labels	9am	11am	6pm	11pm	Grand Total
EXIT	6	82	50	26	164
INCOMING	14	66	15	60	155
Grand Total	20	148	65	86	319

 

HTML Result

 

 

 

The initial design of the counter using Terabee Evo 64px is still ongoing. Making use of python in HTML allows data to update constantly when people are walking past the sensor. Also, since Evo 64px is operating at its fast mode, multiple results are being produced every millisecond. Further improvement can be done by sending the data through the cloud to satisfy IoT requirements. Simple graphics and GUI can be design through HTML for vision purposes.

 

 

 

 

 

Terabee People Counting L-XL

Of course, other than using Terabee Evo 64px to monitor the number of people entering and exiting a confined place like a shopping mall, Terabee People Counting L-XL can also help to count with more accurate data (~98%). The algorithm for People Counting L-XL can count multiple people in both directions at the same time. Next, the algorithm allows a static object to be under the sensor itself.

Robot Motion Monitoring

This sensor is a lightweight form factor that only weighs 12g, hence it can be mounted anywhere at the joints of the robot to detect obstacles from its surrounding. Once obstacles are being detected, Terabee Evo 64px will send a signal to warn the robot to change the path of trajectory to its destination. A relevant case study will be done in the future.

 

Reference

1. M. Hansard, S. Lee, O. Choi, R.P. Horaud, Springer (2012). Time-of-flight cameras - principles, methods and applications. Springer London Ltd.
2. Role of ToF cameras in People Counting - People Traffic Counting. Terabee. (2021, May 18). https://www.terabee.com/role-of-tof-cameras-in-people-counting/.
3. Time-of-Flight principle: Technologies and advantages. Terabee. (2021, May 17). https://www.terabee.com/time-of-flight-principle/.

 

Please contact us if you are interested to see demo of Terabee.