This seed project aims to allow multiple robots to collaborate to transport an object through Multi-Agent Deep Reinforcement Learning. The R&D methodology involves integral utilization of various techniques in deep learning, reinforcement learning, and information science. Deep learning and reinforcement learning will be trained as the controlling algorithm for the multi-robot system. Deep learning model is responsible for processing the input information from the robot sensor, and reinforcement learning algorithm is responsible for training the model from environment feedback. Information science techniques will be used for processing robot sensor input and transforming algorithm output into motion instructions. Together, multiple robots can transport object in synchronization from one location to a designated location.