Client Project
ConsultingSoftware & AI
Autonomous Feed Pushing for an Industrial Loader
As a consulting engagement, we led R&D on autonomous feed pushing for an industrial loader platform, combining a full perception stack with autonomous navigation and edge inference. The project spanned simulation, model evaluation, and hardware-level deployment on Jetson-based compute.
Scope of Work
- Perception stack: object detection, instance & semantic segmentation
- VSLAM and Nav2 integration
- Loader simulation in Isaac Sim
- Vision-Language-Action (VLA) model evaluation
- Edge deployment optimisation
Key Features Delivered
- End-to-end autonomous loader operation
- Isaac Sim environment for motion & task evaluation
- VLA model integration for autonomous decision-making
- Optimised perception & control pipelines for edge deployment
Outcome
The loader started this engagement fully manual. By the end, classical navigation (VSLAM, Nav2) was working alongside vision-language-action models, validated in Isaac Sim and tuned for edge inference on Jetson hardware, running autonomous task execution instead of a joystick.