Tech-fix du jour: AI bike camera alerts cyclists to bad drivers

Ars Technica reports on a AI-powered project for safer cycling.

For $ 399.00 USD you get a small 330 gram camera, powered by a Raspberry Pi Compute Module 4, and equipped with “AI-powered predictive analytics” that you can install under your bike saddle. Facing to the rear, it promises to constantly monitor for bad drivers and give off a audible and visible cue when it detects a potentially dangerous situation. It also visibly signals to upcoming drivers that there is a bike ahead.

This is the pitch on the website:

Copilot is an AI-powered bike light that uses smart sensors to constantly watch the roadway. Cyclists will receive audible and visual feedback from Copilot to indicate driver behavior and warn cyclists of potentially dangerous situations. Audible and visual cues will also indicate to drivers that a cyclist is near in order to avoid potential crashes.


I find this an interesting case of well-meaning techno-fix. It attempts to make cycling safer in countries where there isn’t a tradition of cycling mobility. Good for cities, good for peoople, good for the envirnment. Who could possibly oppose to this? In the Netherlands cycling is ubiquitous and hence car drivers are fortunately much more aware of bikes. Moreover, an increasing number of Dutch cities are now imposing 30km/h speed limits to make traffic safer and more accessible for everyone. But – as our Dutch ‘cycling professor‘ always emphasizes – this is the outcome of political struggles. It is not some kind culturally specific historically grown ‘path dependency’ that others cannot copy.

Hence, I consider these kinds of well-intended ‘solutionist’ smart technologies as backfiring and unhelpful in the end. It puts the onus on individual cyclists to install such a camera. It simply optimizes an existing system of mobility that is car-centric. It depoliticises broader discussions about “the right to the city”, instead of making urban mobility itself the subject of political controversy and debate. For me, the big question with these urban AI applications is how we can move towards more civic uses of urban AI?

Link to article on Ars Technica >>

Leave a Reply