Autonomous Surface Vehicle - Navigation and Obstacle Avoidance
Project Description
This project focused on developing a navigation and obstacle avoidance system for autonomous surface vehicle operating in dynamic open-water environments. The primary challenge was enabling reliable waypoint navigation while maintaining the ability to respond to unexpected obstacles in real time.
We worked within the constraints of an embedded system, where computational efficiency and robustness were critical due to environmental uncertainty such as wind, waves, and moving objects.
Navigation Approach
The system used a real-time waypoint navigation strategy in which the vehicle continuously updated its heading toward an active target location. A feedback controller translated the angular difference between the vehicle's heading and the desired bearing into steering commands.
Waypoints were managed in sequence, allowing the vehicle to progress through a planned route while maintaining stable directional control.
Obstacle Detection and Avoidance
A vision-based perception pipeline provided real-time information about potential hazards in the vehicle's environment. This data was used to determine whether an object posed a collision risk based on its position and motion relative to the vehicle.
When a valid threat was detected, the system temporarily modified its navigation behavior to prioritize avoidance before returning to the original route. This allowed the vehicle to safely navigate around obstacles while preserving overall mission progress.
Results and Validation
The system was first validated in a controlled dry test environment using video-based simulations of boats and obstacles. The full pipeline of computer vision detection, hazard classification, hazard classification, and rudder control, were evaluated and successfully triggered correct avoidance behavior.
The system was then deployed in open-water conditions, where it successfully detected real obstacles (such as docks), generated avoidance waypoints, and rerouted the vessel without collision. This demonstrated end-to-end functionality of the navigation and obstacle avoidance stack in real-world conditions.
Conclusion
This project strengthened my understanding of real-time robotic navigation in uncertain environments and reinforced the importance of designing robust, lightweight decision-making logic for embedded autonomous systems.
Skills
Autonomous Systems and Control
- Waypoint navigation
- Feedback control
- Real-time steering logic
Perception-to-Action Systems
- Vision-based hazard integration
- Decision-making logic
- Filtering and threshold tuning
Robotics Engineering
- Autonomous surface vehicle behavior
- Real-world testing
- System-level integration
Create Your Own Website With Webador