IndiGO is Bluecity.ai’s computer vision and traffic data platform. It provides real-time, multi-modal traffic data and analytics through the use of artificial intelligence and LiDAR-based sensors.
To prove the accuracy of the solution, the company has been performing a series of ongoing studies. Having confirmed that the system accurately detects the number of cars passing through an intersection in a previous study, the Bluecity.ai team is now addressing speed. This latest study seeks to understand how accurate IndiGO is in measuring the speed of road users.
The study took place in Kelowna, British Columbia at an intersection where an IndiGO 3D sensor had been installed. This intersection had been used in the previous studies on counting vehicles and is now being used to determine accuracy of speed detection.
Speed would be measured as cars moved across a predefined area referred to as the red zone. The size of the red zone was extracted from LiDAR data and matched with Google Satellite images to make sure the distance travelled matches with reality.
For each vehicle entering and leaving the red zone, time stamps were registered. The traffic being measured travelled in one of four directions: North to South, South to North, East to West and West to East.
Speed was measured based on the location of the vehicle with a 10 frame per second analysis. The study generated three types of speed values:
Instantaneous Speed (IS) or current speed, was based on the location of the car at the current and previous frame.
The Moving Average Speed (MAS) was the average of the vehicle’s instantaneous speed.
The Zone Speed (ZS) was calculated based on the stime stamp of the car entering and leaving the predefined red zone (the distance travelled).
As with earlier studies, the data processed by the AI algorithm would be compared with the ground truth. The ground truth speed would be measured manually by an analyst who watched the recorded LiDAR data. In this study, the ground truth speed of 480 cars was measured and compared with the speed reported by the LiDAR sensor. It should be noted that the amount of traffic travelling North and South was significantly less than the traffic travelling East and West, which was about 90%. A Mean Absolute Percentage Error (MAPE) was used to measure the average error over all the samples.
The results showed an overall accuracy of 92% in measuring speed using the LiDAR sensor. The 92% included measurement of both zone speed and moving average speed for cars travelling under 35 kilometers per hour in all directions.
The results showed an overall accuracy ofin measuring speed using the LiDAR sensor. The 92% included measurement of both zone speed and moving average speed for cars travelling under 35 kilometers per hour in all directions.When cars were travelling more than 35 km/hr the overall accuracy was slightly higher at 96%.
This slight difference in accuracy is due to the fact that cars going through the intersection at more than 35 kilometers per hour are moving at a more constant speed (i.e., they did not have to stop for red lights). As a result, the sensor picked up less noisy data and the accuracy was higher.
With these results on speed measurement confirmed, there are many ways in which cities, engineering and consulting firms can use IndiGO to enhance planning. Studies on driver behaviour, and how to monitor over speeding are possible.
The effects of weather and lighting on speeding are also particularly interesting because IndiGO sensors function in all types of lighting and extreme weather conditions. Additionally, traffic light timing can be studied to improve efficiency and reduce traffic congestion.
The Bluecity.ai team is comitted to ongoing studies that will explore performance in variable weather and lighting conditions, vehicle classifications, safety analytics and more. The company’s goal is to make streets safer for pedestrians, cyclists and motorists. We envision a world where traffic accidents are rare, driving to work is a pleasure and carbon emissions are reduced. Make sure to read our next studies on the accuracy, reliability, and broader insights available from IndiGO sensors.