Image Image Image Image Image Image Image Image Image Image

Bartlett School of Architecture, UCL

Scroll to top




CuGo is an interactive board game, where human and robot players collaborate to achieve a shared goal. It consists of a game board, on which modular reconfigurable robots, perform unpredictable movements, as well as passive blocks that are moved by both human and robot players. The goal of the game is to stack the passive blocks as high as possible.The player moves the passive blocks on the board into a position where they think the robot can grab them. When the passive blocks are in reachable zone, the robots use their inverse kinematic intelligence to reach the passive units, but when they are not accessible, the robots suggest alternative positions to the player.

Dramatic developments in artificial intelligence have established the fact that the performance of AI for specialised tasks will surpass those of humans. CuGo is a tangible platform where both intelligence of human and machine are encouraged to participate to achieve a common goal, breaking the anxiety between two entities. While most artificial intelligence studies focus on autonomous systems, CuGo seeks a comprehensive understanding of artificial intelligence, proposing a human centric, shared autonomy.

CuGo introduces a user-friendly environment not only for technologists, but for people without any AI related knowledge. With a proper learning model, CuGo creates a mutual learning system where players can observe how the robot performs and learn the logic behind their movements, at the same time, the robots can gradually build their “personality”, namely unique ways of moving. The current gaming process is a basic demonstration of the CuGo system, future variants of CuGo with a further developed learning ability could provide more freedom to play, allowing human players to design their own rules and goals for the game.


This is the form development of CuGo’s active unit. It has angular rotational axis which makes it more complex and graceful motions than perpendicular axis. Single module can rotate right (120 degrees) middle (0 degrees) and left (-120 degrees)


Diagram above shows how to interact with the active robots. The goal of the game is to stag as much passive units as possible.



The 9 by 9 board provides power to the active units via conductive surfaces, and The passive units, which are the targets of the active units, has ArUco markers to provide three-dimensional position data to active units.


ArUco markers are tracked via Logitec Webcam

The exploded diagram below illustrates the components inside the robot. The structural material is 3d printed using PLA filaments.

The major aim of the active unit design process was to make it smaller and lighter. The current active unit’s edges are 6 by 6 by 6 cm with 60 grams weight. The smallest of the shelf actuators and control board have been used. In total it costs around 25 GBP per one unit.


While industrial robot arm’s rotation axis are perpendicular to each other CuGo’s axis are not. Therefore, custom DH parameters are assigned for inverse kinematics.


connection types between (left) passive modules and robot – permanent magnets, (middle) robot and robot – interlocking joints, (right) robot and robot – molex cable joints


From 13 cm to 6 cm

Above image is the iterations of active units.

three exploded diagrams of first, second and last iterations.






  • Rus, D., Butler, Z., & Kotay, K. (2002). SELF-RECONFIGURING ROBOTS. Communication of the ACM, 45(3), 39–45
  • Salemi, B., Rus, D., Moll, M., Lipson, H. O. D., Klavins, E., & Chirikjian, G. S. (2007). Modular Self-Reconfigurable Robot Systems, (March)
  • Yim, M., Shen, W. M., Salemi, B., Rus, D., Moll, M., Lipson, H., … Chirikjian, G. S. (2007). Modular Self-Reconfigurable Robot Systems [Grand challenges of robotics]. IEEE Robotics and Automation Magazine
  • Seo, J., Paik, J., & Yim, M. (2019). Modular Reconfigurable Robotics, 1–25