What we have to do is to reproduce this without the wires.
We also have to work on the Bluetooth communication to break the 7 devices limit. One Bluetooth device can either be a master or a slave. The PC sends information about each module position and, as a master, can only control 7 modules at once ("piconet" = small network). A Bluetooth device can change its state from master to slave and vice-versa. You could hence think about a more advanced protocol with slaves turning into masters and relaying the information to other devices ("scatternet" = set of piconets). Rico Möckel is currently working on the "scatternet" protocol, he also designed the FPGA/Bluetooth and electronics parts of the module. With Elmar Dittrich who was in charge of the mechanical design, they managed to find the cheapest pieces for the module without sacrifying the robustness of these elements.
In theory, the scatternet idea could work with up to 60 devices but there are many other aspects like interferences or realtime constraints. We guess that it will be hard to work with more than 30 devices. First we would have to cope with the problems mentionned above. Another problem could be related to the limited torque provided by the servos. If the robot is too heavy, the motors won't be able to move. Adding modules will bring more errors, it could be hard to have a stable robot which won't fall apart. We also want to redesign the control application with a nice GUI in Java. The whole system will work in Java and we use the free Rococo Java stack for Linux. A Bluetooth Stack is a kind of driver which allows interaction between the Java API and the Linux driver (BlueZ). The information are sent through a Bluetooth USB stick, we use the RFCOMM protocol (serial communication) to communicate with the module.
We are now assembling other modules to test the system with more units and reproduce the examples shown in the videos (see links below).
More info at :