A large part of our research is done in the domain of robotics. For this we have several robots available.
The Oncilla platform is a light-weight bio-inspired quadruped robot. This compliant robot is used for research on legged locomotion on rough terrain.
Tensegrities (Tensile-Integrity) are structures consisting of compressive elements (struts) held together by tensile elements (strings or springs). Buckminster Fuller defined them as islands of compression in a sea of tension.
This design strategy scales from the cellular level up to buildings and bridges and allows for lightweight structures of varying stiffness.
Because class-1 tensegrities are free of joints (no connected struts), it becomes possible to build highly compliant robots which can exhibit large shape changes (e.g. Figure below).
Our tensegrity platform combines rich sensor integration with a lightweight mechanical design (sub kg for the configuration shown below).
Applications of these robots range from morphological computation (the body as a brain) to space exploration (satellites, landers).
In our group we also have a dog robot, which was originally designed during a student hardware design project.
Our master thesis students developed a new low-cost (<70 EUR/robot), easy-to-build swarm robotics platform. The current setup consists of 20 robots with accelerometer, short and long distance sensors, wireless communication, 2 mouse sensors for precise odometry, extension ports... These robots are mainly used for research into Reservoir Computing based multi-agent reinforcement learning.
In the lab we also have an inverted pendulum setup available. This classic setup is used to test underactuated control methods and for learning inverse dynamics. The platform was designed during a student hardware design project.
These include e-puck robots, a Bioloid Kit and Webots licenses. The e-puck robots are extended with a long-range distance turret which was developed in our lab during a student project. Many other robots were bought, build or modified for use in student hardware and software projects including a segway robot, a helicopter and a robot car.
Finally another collection of robots was built - mainly by Francis wyffels and Michiel D'Haene - using lego and the Dwengo microcontroller platform for educational purposes. This includes 6 autonomous robot cars, some demos for the WELEK robot competitions, a robot illustrating the concept of morphological computation and a crawling robot which learns moving forward by itself using reinforcement learning (Q-learning) of which you also can see a movie.