Limited mobility, scalability, and environmental adaptation are some of the major issues facing current modular robotics systems. In fact, current systems' potential in practical applications is limited because they are still bulky, depend on separate modules with limited degrees of freedom, and lack large-scale simulations. In order to address these problems, Project M.I.R.A.S. presents a compact, self-assembling modular robotic system that draws inspiration from insect swarms. Its key component is the worker module, that features three DS-M005 servo motors and worm-gear mechanisms for steady 3D movement and high torque. Furthermore, energy distribution is achieved through a shared grid system that passes continuous current, managed by thermal sensors and MOSFETs to prevent overheating. For energy-efficient connections, neodymium magnets and electromagnets are used, forming an “energize-to-release” system, which allows for reliable attachment and detachment between modules. A key innovation of the project are the voxelization algorithms, which underpin the system’s scalability. With the help of these algorithms, the modules can effectively recreate complex 3D structures, resulting in faster setups and task-specific adaptability. Complementing this, deep reinforcement learning is employed to optimize movement. By utilising TensorFlow and OpenAI Gym, neural networks are trained to dynamically adjust motor angles in order to achieve efficient forward locomotion by maximising x-axis movement in a 30-second performance metric. To enhance user interaction and prototyping, I developed a VR application using Unity, which was then tested on the Quest 3 headset. By including passthrough and hand-tracking capabilities, these simulations enable users to handle individual modules and prototype structures in a natural manner inside real-world settings. Additionally, scalability is further achieved by Zigbee mesh networking, which provides fault-tolerant, low-latency communication across thousands of modules. At last, with in-place 3D-printing, problems such as center-bearing load stability were fixed. To sum up, the end product is a modular robotic system with uses in industrial automation, planetary exploration, and catastrophe recovery that can construct stable, scalable configurations in 3D. On top of that, by addressing labour shortages and offering modular solutions where continuous adaptation is required, Project M.I.R.A.S. has the potential to transform the role of robots in a range of industries.