Vision and Off-road Systems

How it works

Driver assistance systems for industrial vehicles, such as forklifts and aerial work platforms, are based on advanced sensor technologies and local data processing (edge computing). They use video, radar, or LiDAR sensors to capture real-time information about the vehicle’s surroundings, both indoors and outdoors. Using machine learning algorithms – Multilayer Perceptron (MLP), Support Vector Machine (SVM) – and deep learning (DL), such as Convolutional Neural Networks (CNN) or Recurrent Neural Networks (RNN), the data is analyzed to detect and classify objects around the vehicle. The system alerts the driver to potential hazards via a man-machine interface (MMI) in the form of visual or audible signals. This assistance anticipates accident risks and can even suggest corrective actions. Additional dashcam-like functionality for recording pre- and post-impact sequences can also be provided to provide contextual information in the event of an incident and enable corrective action to be taken.

Managing Complex Environments

One of the major challenges in image processing is the management of indoor and/or outdoor environmental conditions, such as limited viewing angles, occluded objects, or variations in brightness (day – night) and weather conditions (fog, rain, etc.). To ensure reliable scene reconstruction even under difficult conditions, systems can combine:

• Multiple sensor technologies (camera, radar, LiDAR), whose network of captured data forms a huge dataset that can be used for in-depth analysis by artificial intelligence.
• Video sensors using advanced technologies such as panoramic vision, 3D stereoscopic vision and infrared.
• Robust image recognition and classification algorithms.

These technologies provide optimal support for the operator while ensuring increased safety in complex work situations.

Use cases

• Forklift trucks: By analyzing the environment around the vehicle, vision systems assist the operator in handling the forklift, which is even more delicate in very narrow aisle (VNA) intralogistics environments. In addition to avoiding potential collisions, cameras or scanners facilitate the operator’s work by helping him to position the forks correctly to pick up or put down a pallet, even at very high heights (> 10 m), without risk of falling.
• Aerial work platforms: Vision systems allow the platform to operate in poor visibility conditions, extending working hours while maintaining safety.

By preventing collisions and facilitating maneuvering, they greatly reduce the likelihood of personal injury and property damage, increase productivity, and reduce operator fatigue and stress.

Key points for the integration of ADAS equipment in industrial off-road vehicles

Although these applications do not necessarily require the development of complex algorithms, several aspects deserve special attention:

• Integration into the final vehicle architecture: The system must be compact enough to be easily integrated into the vehicle chassis. This is even more important in a retrofit project where space is limited.
• Understanding the environment in which the industrial vehicle operates: An industrial environment requires a solution that is rugged enough to withstand vibration, temperature changes, and possible shocks to minimize the risk of failure. In addition, the solution must adapt the choice of sensors and algorithms to the vehicle’s environment, whether indoors and/or outdoors.
• Edge Computing Constraints: In the case of local processing, power, thermal, and ruggedness constraints are essential criteria for system design and integrated SoC selection. These criteria determine the overall performance and reliability of the deployed devices.