Exeros is paving the road to smarter transportation. Expanding the understanding of camera tech and what constitutes risky scenarios for fleet drivers, the company is one step closer to achieving its vision of making sure that every road user makes it home safely each day.
How it began
Exeros have been in the business for over 12 years, deploying their well-established fleet CCTV systems to some of the biggest fleets in the UK. The company uses edge AI to recognize certain risks on the road such as collisions, lane departure, driver fatigue, mobile phone use and prolonged distraction, then alerting the driver of such events.
Fleet customers face significant risks on the road every day with human error accounting for 94% of all accidents worldwide. In a recent survey, 9 out of 10 lorry drivers admitted to streaming video content on mobile devices whilst driving long-haul.
With aims of reducing new risks to fleet operators, Exeros focused on ramping up the quality and quantity of the events its cameras can recognise by training new AI ADAS algorithms.
The tech
Exeros have recognised that the AI algorithms that they deploy are very good for direct reduction of fleet risk. However, they have acknowledged there are other edge road scenarios that might affect road risk and improve ADAS systems.
Looking at expanding what the cameras can see by training new sets of AI algorithms to detect new street furniture such as road signs and road crossings and other animate or inanimate objects on the road including pedestrians.
More specifically, Exeros has been looking at testing the ability to read road signs as well as looking to improving image recognition of partially obscured signs, be it by overgrown vegetation or otherwise.
Exeros has also been a step further by testing different speed-based accuracies as well as improving the image recognition in different lighting scenarios. All the validation tests with the CAM Scale-Up initiative are aimed at securing ground truth data against which ADAS accuracy can then be benchmarked and measured.
Testing
Exeros has run specific tests using the highly surveyed environments of the testbeds:
The first strand of work was all about establishing ground truth data and improving ADAS accuracy. SMLL provided the company access to 5km of 3D models where they were able to tag certain types of signs and assets. This allowed Exeros to train their algorithms off these known locations.
Exeros also wanted to explore how accuracy can be improved in different scenarios by analysing the repeatability of visual circumstances. At UTAC, Exeros was presented with an on-track urban testing environment simulating real-world scenarios. Thanks to the repeatability of specific situations, Exeros aimed to fine-tune its AI algorithms to achieve over 99% accuracy.
Challenges
Exeros had ensured that, as well as the necessary hard ADAS improvements, the exposure of participating in the CAM Scale-up initiative would help to secure further relationships for its improved ADAS and fleet data solution. It already has a number of companies interested and the initiative should serve as the commercial starting pistol for what promises to be an enhanced fleet safety digital eco-system.
For more on this check out Exeros Technologies case study over on Zenzic.
Stay tuned on what’s to come. For further info on Exeros, contact us here.
