{"id":184,"date":"2025-09-07T19:41:25","date_gmt":"2025-09-08T02:41:25","guid":{"rendered":"https:\/\/blogs.ubc.ca\/ajaysharmaportfolio\/?page_id=184"},"modified":"2025-09-08T00:24:41","modified_gmt":"2025-09-08T07:24:41","slug":"ubc-subbots","status":"publish","type":"page","link":"https:\/\/blogs.ubc.ca\/ajaysharmaportfolio\/ubc-subbots\/","title":{"rendered":"UBC Subbots"},"content":{"rendered":"

 <\/p>\n

\n

\"\"The UBC Subbots team focused on advancing the performance and autonomy of an Autonomous Underwater Vehicle (AUV) through innovations in machine learning, navigation algorithms, and hardware-software integration. The work included developing high-accuracy object detection models trained on both public and custom datasets, implementing path planning and trajectory optimization algorithms, and prototyping custom circuits for orientation tracking sensors. These combined efforts resulted in a system capable of improved target detection, more efficient navigation, and enhanced positional awareness, all of which are critical for success in underwater robotics competitions and real-world marine applications.<\/p>\n

Introduction<\/h3>\n

Underwater environments pose unique challenges for autonomous systems due to limited visibility, unpredictable currents, and constraints in communication. The UBC Subbots team addresses these challenges by designing, building, and programming an AUV capable of performing tasks independently during competitions such as RoboSub. The contributions in this project aimed to increase detection accuracy, reduce navigation time, and improve vehicle stability. A holistic approach was taken, combining machine learning techniques with navigation control algorithms and sensor hardware integration, ensuring that the system performs reliably under demanding underwater conditions.<\/p>\n

Machine Learning and Object Detection<\/h3>\n

Accurate and robust object detection is a core requirement for the AUV to identify mission elements such as gates, buoys, or markers. Models were trained using both the COCO dataset and a custom underwater dataset collected and labeled by the team. The COCO dataset provided a strong foundation of general object recognition, while the custom dataset addressed underwater-specific challenges like low visibility, image distortion, and color loss at depth. The integration of these datasets improved model robustness and significantly increased detection accuracy.<\/p>\n

In addition to dataset development, the models were optimized to run efficiently on the onboard processing unit, balancing accuracy and inference speed. This allowed the vehicle to make real-time decisions during missions without exceeding hardware limitations. By reducing false positives and improving detection rates, the AUV became more reliable in task execution and environmental awareness.<\/p>\n

Path Planning and Trajectory Optimization<\/h3>\n

The next major focus was the design and implementation of path and trajectory generation algorithms. These algorithms used visual input from the detection system to calculate optimal paths toward mission targets. Special attention was given to reducing unnecessary movement and optimizing trajectories to minimize navigation time.<\/p>\n

Simulation results suggested that the newly implemented algorithms could reduce navigation time by as much as 25% compared to baseline methods. This efficiency gain is particularly important in competitions where the AUV must complete a sequence of timed tasks. Beyond competitions, such optimization has practical applications in reducing energy consumption and extending operational endurance for underwater research missions.<\/p>\n

Sensor Circuit Design and Prototyping<\/h3>\n

\"\"To improve orientation and positional tracking, the project included the design and prototyping of custom circuits for gyroscope and magnetic field sensors. The gyroscope circuit provided accurate angular velocity data, while the magnetometer circuit allowed the vehicle to estimate heading relative to Earth\u2019s magnetic field. Together, these sensors improved the accuracy of the vehicle\u2019s Inertial Measurement Unit (IMU).<\/p>\n

The circuits were carefully designed to minimize electrical noise and ensure reliable data transmission in underwater conditions. Testing confirmed improved responsiveness and stability, allowing the vehicle to execute complex maneuvers such as turning, maintaining heading, and compensating for disturbances.<\/p>\n

Hardware-Software Integration<\/h3>\n

\"\"The strength of the project lay in the integration of hardware and software solutions. Data from machine learning models, navigation algorithms, and orientation sensors were fused together to generate an accurate real-time estimate of the AUV\u2019s position and orientation. This fusion enabled the AUV to react dynamically to its environment and maintain control despite external disturbances.<\/p>\n

Hardware integration included designing mounts and waterproof enclosures for the new sensors, ensuring reliable operation in submerged conditions. Software integration focused on synchronizing data streams from multiple sensors and aligning them with the control system. This allowed the AUV to achieve smooth navigation, precise positioning, and stable operation during tests.<\/p>\n

Results and Discussion<\/p>\n

The project successfully enhanced the AUV\u2019s autonomy, navigation efficiency, and sensor reliability. Object detection accuracy was significantly improved, ensuring that mission-critical elements could be consistently identified. The trajectory optimization algorithms demonstrated measurable reductions in navigation time, supporting faster and more efficient mission execution. Orientation tracking circuits enhanced stability, enabling the vehicle to handle underwater tasks more reliably.<\/p>\n

Challenges encountered included the computational limitations of onboard processors, the need for extensive dataset refinement, and calibration difficulties in variable underwater conditions. Future improvements may involve migrating to more powerful hardware for real-time processing, expanding the custom dataset with diverse underwater environments, and further refining sensor calibration procedures.<\/p>\n

Conclusion<\/h3>\n

The UBC Subbots project demonstrated a comprehensive approach to advancing underwater vehicle autonomy by combining machine learning, trajectory optimization, sensor hardware design, and system integration. These efforts not only improved the AUV\u2019s competition performance but also provided a strong foundation for real-world marine research applications, such as environmental monitoring, search and rescue, and underwater inspection. With continued refinement, the vehicle can evolve into a fully autonomous underwater platform capable of operating in complex and unpredictable marine environments.<\/p>\n

 <\/p>\n","protected":false},"excerpt":{"rendered":"

  The UBC Subbots team focused on advancing the performance and autonomy of an Autonomous Underwater Vehicle (AUV) through innovations in machine learning, navigation algorithms,…<\/p>\n

Continue ReadingUBC Subbots<\/span><\/a><\/div>\n","protected":false},"author":103569,"featured_media":0,"parent":0,"menu_order":0,"comment_status":"closed","ping_status":"closed","template":"","meta":{"footnotes":""},"class_list":["post-184","page","type-page","status-publish","hentry","entry"],"_links":{"self":[{"href":"https:\/\/blogs.ubc.ca\/ajaysharmaportfolio\/wp-json\/wp\/v2\/pages\/184","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/blogs.ubc.ca\/ajaysharmaportfolio\/wp-json\/wp\/v2\/pages"}],"about":[{"href":"https:\/\/blogs.ubc.ca\/ajaysharmaportfolio\/wp-json\/wp\/v2\/types\/page"}],"author":[{"embeddable":true,"href":"https:\/\/blogs.ubc.ca\/ajaysharmaportfolio\/wp-json\/wp\/v2\/users\/103569"}],"replies":[{"embeddable":true,"href":"https:\/\/blogs.ubc.ca\/ajaysharmaportfolio\/wp-json\/wp\/v2\/comments?post=184"}],"version-history":[{"count":6,"href":"https:\/\/blogs.ubc.ca\/ajaysharmaportfolio\/wp-json\/wp\/v2\/pages\/184\/revisions"}],"predecessor-version":[{"id":292,"href":"https:\/\/blogs.ubc.ca\/ajaysharmaportfolio\/wp-json\/wp\/v2\/pages\/184\/revisions\/292"}],"wp:attachment":[{"href":"https:\/\/blogs.ubc.ca\/ajaysharmaportfolio\/wp-json\/wp\/v2\/media?parent=184"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}