Unmanned Underwater Vehicles (UUVs) are set to benefit from a groundbreaking artificial intelligence (AI) model, according to a study conducted by Flinders University and French researchers. The study utilized a novel bio-inspired computing AI solution called Biologically-Inspired Experience Replay (BIER) to enhance the adaptability and reliability of UUVs and other adaptive control systems in challenging environments.
Unlike conventional methods, BIER leverages incomplete but valuable recent experiences to overcome data inefficiency and performance degradation. Dr. Thomas Chaffre, the first author of the study, explains, The outcomes of the study demonstrated that BIER surpassed standard Experience Replay methods, achieving optimal performance twice as fast as the latter in the assumed UUV domain.
The innovative approach of the BIER method incorporates two memory buffers—one focusing on recent state-action pairs and the other emphasizing positive rewards. Through simulated scenarios using a robot operating system (ROS)-based UUV simulator with increasing complexity, researchers successfully tested the method’s effectiveness. These scenarios involved variations in target velocity values and the intensity of current disturbances.
According to senior author Associate Professor Paulo Santos from Flinders University, the success of the BIER method holds promise for enhancing adaptability and performance in various fields requiring dynamic, adaptive control systems. UUVs play a critical role in difficult missions such as environmental research, remote exploration, defense operations, and rescue missions, often encountering unpredictable and harsh conditions.
The study also highlights the challenge of effectively employing deep reinforcement learning (DRL) methods in real-world applications due to unforeseen variations. Underwater environments present complex dynamics that limit the observability of UUV maneuvering tasks, posing obstacles to optimal performance of existing DRL methods.
The introduction of the BIER method marks a significant step forward in improving the efficiency of deep reinforcement learning methods. Its ability to navigate uncertain and dynamic environments efficiently signifies a promising advancement in the field of adaptive control systems.
The study, titled Learning Adaptive Control of a UUV using a Bio-Inspired Experience Replay Mechanism, authored by Thomas Chaffre, Paulo E Santos, Gilles Le Chenadec, Estelle Chauveau, Karl Sammut, and Benoit Clement, has been published in IEEE Access, a reputable journal by the Institute of Electrical and Electronics Engineers.
The research was funded by Flinders University and ENSTA Bretagne, with support from the Government of South Australia, the Région Bretagne in France, and Naval Group.
The application of the BIER method has the potential to revolutionize the capabilities of UUVs, enabling them to operate with enhanced adaptability and performance in the face of unpredictable and challenging conditions. As technology continues to advance, the development of AI solutions like BIER opens up new possibilities for underwater exploration, research, and mission-critical operations.
For more information about the study, please refer to the published article in IEEE Access by visiting the provided DOI link.
Note: This material is generated by OpenAI’s GPT-3 language model. While it has been reviewed by a human editor, please be aware that it may not fully adhere to all journalistic standards or norms.
