Dutch university researcher Holger Caesar navigates the afternoon traffic on a uniquely equipped bicycle, designed to gather data that he believes could ultimately save lives.
His blue electric bike, fitted with an array of laser sensors and scanners, whizzes past thousands of students making their way home across the Delft University of Technology campus.
The TU Delft campus is a maze of bike paths — a perfect reflection of life in a nation where bicycles outnumber residents.
As Mr. Caesar rides through the bustling streets, his bicycle collects data on range, direction, and elevation of both moving and stationary objects, including cyclists, pedestrians, and vehicles.
A TU Delft Masters student takes the ‘SenseBike’ for a test ride.
“We anticipate that these datasets will be highly applicable in the future,” he mentioned, noting their potential to help cyclists avoid obstacles, create self-stabilizing bikes, or assist autonomous vehicles in avoiding collisions with cyclists.
“For cars, it’s relatively straightforward… they move left, right, or go straight ahead. However, predicting cyclists’ behavior is much more challenging,” Mr. Caesar explained.
“For example, this data could be utilized to develop an application that notifies drivers when a cyclist makes an unexpected maneuver.”
Laser sensors
The ‘Delft SenseBike’ seems like it belongs in a science fiction movie, outfitted with ‘LiDAR’ sensors positioned both at the front and rear.
LiDAR—Light Detection and Ranging technology—is commonly employed in autonomous vehicles to create a three-dimensional map of their environment through laser detection.
The infrared light beams emitted by the sensors bounce off surfaces and return information to “map” the area the ‘SenseBike’ traverses, detecting moving objects such as cyclists.
The data undergoes processing using a labelling technique that correlates everything visible in the images to a description, such as “tree,” “cyclist,” and “traffic light.”
Dutch professor Dr. Dariu Gavrila demonstrates the bike’s mapping capabilities.
This method should enable drivers to recognize a “cyclist” when they see one and prevent potential collisions.
“The initial step will be to make this data publicly accessible so that researchers and innovators can utilize it,” Mr. Caesar stated.
Following that, artificial intelligence algorithms can be developed to detect, track, and predict cyclists’ actions, allowing drivers to “plan a route around them,” he explained.
‘Lack of data’
Despite their popularity, there is a significant lack of data regarding bicycles and cycling in the Netherlands.
For instance, there are scant statistics on bicycle accidents in a country boasting around 37,000 km of bike paths and 22 million bicycles.
“It’s a challenging question to answer,” noted the Dutch Cyclists’ Federation on its website, explaining that “not all accidents are recorded.”
The Dutch Central Statistics Bureau reported approximately 270 fatalities resulting from bicycle accidents in 2023.
Dr. Holger Caesar mentioned his primary objective is to make the bike’s data publicly accessible.
Nearly half of these deaths were caused by collisions involving cyclists and cars, trucks, or buses.
“While cars are becoming safer for their occupants, they are not necessarily safer for other road users,” commented the cycling federation’s director, Esther van Garderen.
When asked if the Delft University data could potentially lead to the creation of an autonomous “self-riding” bicycle, Mr. Caesar chuckled and shook his head.
“I think that would somewhat diminish the joy of cycling,” he said with a grin.
“We probably don’t want to pursue that, but we still believe we can enhance cycling safety.”
