Voronoi-Based Multi-Robot autonomous exploration is, indeed, an effective method for mapping and navigating new areas. However, traditional methods use fixed rules that can fail in complex situations. In contrast, deep reinforcement learning (DRL) offers a better option, as it allows robots to learn from experience. Consequently, this literature review discusses advanced Voronoi-Based Multi-Robot exploration with DRL, specifically focusing on environment representation, reward function design, and coordination among robots. Notably, Voronoi diagrams help in dividing areas for exploration, while DRL enables robots to adapt to surprises and enhance their actions.
Various research studies have delved into numerous deep reinforcement learning (DRL) structures and training techniques designed specifically for Voronoi-Based Multi-Robot exploration. Some researchers advocate for a centralized training strategy coupled with the deployment of individual robots, while others advocate for a decentralized training method. There are significant challenges to address, including the development of a reward system that not only incentivizes exploration but also prevents collisions and fosters collaboration between robots. Furthermore, the representation of the environment plays a critical role in the overall success of the exploration process. Popular methods of representing the environment include grid-based maps and occupancy grids.
In spite of the advancements, researchers still face numerous challenges with Voronoi-Based Multi-Robot autonomous exploration using DRL. These challenges include the high computational expense of training DRL agents in large state spaces, difficulties in tapping into learned policies across different environments, and the need for robust communication and coordination mechanisms. Some future research directions involve exploring more effective DRL algorithms, developing more expressive environment representations, and incorporating prior knowledge into learning. Besides, researchers must address safety concerns and ensure that Voronoi-Based Multi-Robot systems can be made useful in realistic environments, which remain huge areas for investigation.
Click here to get the complete project:
