Autonomy for Off-Highway Vehicles and Machinery
Helping Off-Highway OEMs and Tier 1 suppliers develop autonomous products with AI safety in place.
Get in touchAI Safety and Perception Systems for Industrial Autonomy
We bring more than 10 years of experience in automotive-grade computer vision, perception, and safety engineering to off-highway autonomy applications.
-
AI Safety
We develop safety-oriented perception systems using FuSa, safety analysis, and AI safety engineering for off-highway vehicles and machinery in demanding environments.
-
AI Perception
Computer vision and machine learning solutions for environmental awareness, scene understanding, and autonomous decision support.
-
Object and Hazard Detection
Detection of people, vehicles, machinery, obstacles, and other relevant hazards in complex operating environments.
-
Sensor Fusion
Researching and deploying the correct mix of sensors to improve performance of vehicles and machines in harsh environments.
-
Object Tracking
Reliable tracking of dynamic and static objects over time to support stable behavior in real-world industrial scenarios.
Off – Highway Industry name use cases
Autonomy, AI Safety and Perception AI solutions for the Off-Highway and Machinery industry.
Decrease production cost for autonomous lawn mowers
Autonomous lawn mowers need to operate reliably in changing outdoor conditions while remaining cost-efficient for large-scale production. Overly complex sensor architectures can increase hardware costs, processing requirements, and integration complexity without necessarily delivering proportional improvements in performance. Manufacturers therefore need to find the right balance between perception accuracy, robustness, and overall system cost.
Sensor fusion with cameras and LiDAR
A cost-effective approach is to combine cameras and LiDAR in a carefully optimized sensor setup rather than relying on a more complex or over-specified architecture. Cameras provide rich visual information for object recognition, boundary detection, and scene understanding, while LiDAR adds accurate distance measurement and spatial awareness. Fusing these inputs improves obstacle detection, navigation, and reliability in challenging outdoor conditions such as glare, shadows, and uneven terrain. The perception pipeline can then be optimized for the target hardware to reduce computational requirements and support efficient embedded deployment. The objective is to achieve the required autonomous performance with the simplest and most cost-effective combination of sensors, hardware, and perception software.
- Development of Autonomous Features
- Computer Vision / Machine Learning
- Sensor Fusion
- Perception AI
- Embedded AI Optimization
- R&D
Outdoor environments create unpredictable perception challenges
Autonomous outdoor machines must operate across changing lighting, weather, terrain, and vegetation conditions. Shadows, direct sunlight, wet surfaces, tall grass, partially hidden objects, and seasonal changes can significantly affect perception performance. A system that performs well in controlled testing may therefore struggle when deployed across thousands of real-world environments.
Robust perception trained for real-world variability
Computer vision and machine learning models can be developed and validated using datasets that represent the full range of environments the machine is expected to encounter. This includes difficult lighting conditions, different types of terrain, partially occluded obstacles, people, animals, garden furniture, and other relevant edge cases. Data analysis and targeted model development help identify weak areas in the perception system before deployment. Models can then be optimized and validated against clearly defined operational scenarios to improve consistency across changing outdoor conditions. This results in a perception system designed around real-world variability rather than idealized test environments.
- Computer Vision / Machine Learning
- Dataset Development and Analysis
- Perception AI
- Object Detection and Classification
- AI Validation
- R&D
Autonomous machines must navigate safely around dynamic obstacles
Autonomous lawn mowers and other outdoor robots operate in environments where people, pets, toys, tools, vegetation, and other obstacles can appear unexpectedly. The system must distinguish between traversable terrain, boundaries, and objects that require avoidance while continuously planning its route. Reliable perception and fast decision-making are essential for maintaining both operational performance and safety.
Real-time perception for safe autonomous movement
A real-time perception pipeline can detect, classify, and track objects around the machine while continuously providing environmental information to the navigation system. Camera and range-sensor data can be fused to estimate object position, distance, movement, and free space more accurately. This enables the autonomous platform to slow down, stop, reroute, or adjust its trajectory when obstacles are detected. The software can be optimized for embedded hardware so that perception and navigation decisions happen locally with minimal latency. The result is more reliable autonomous movement across complex and dynamic outdoor environments.
- Development of Autonomous Features
- Object Detection & Tracking
- Sensor Fusion
- Perception AI
- Computer Vision / Machine Learning
- Edge AI
