_____________________________________________________________________________________________ Bucher Municipal
Computer vision can identify relevant conditions around the sweeping process and SmartAssist can use those inputs to influence machine operation
sweeping. We then deliberately built our own capability through a Knowledge Transfer Partnership with Harper Adams University from 2020 to 2023. From there we moved progressively from research hardware into real vehicles,” he details.
“ In 2024, we publicly demonstrated the SmartAssist concepts at Mercedes-Benz World and then moved into Alpha trials with internal and selected customer operation. During 2025 we established our Command Centre and moved through pre-Beta into customer Beta operation, including vehicles operating across London. We are now expanding that learning across a wider UK Beta fleet.
“ The development process itself is deliberately iterative. We collect representative road and debris data, label and train the models, test them through deterministic replay and controlled testing, integrate them onto the vehicle, and then validate them again in real-world sweeping conditions. The trials have demonstrated that the underlying concept is viable. Computer vision can identify relevant conditions around the sweeping process and SmartAssist can use those inputs to influence machine operation. Our early SmartCSi repeatability work has provided encouraging evidence that the system can distinguish between cleaner and dirtier road conditions and repeatedly identify similar contamination patterns. What the Beta programme is doing now is moving beyond that proof of concept and establishing the confirmations needed for a production product. We are testing repeatability, reliability, operating limits, operator
ccemagazine. com 57