Design and Simulation of a Gravity Compensator with a Noncircular Pulley - Spring Mechanism

The purpose of this master thesis is to analyse the effects induced by gravity compensators in robotic systems. Nowadays robotic research is more focused on attaining energy-efficient and safe solutions. For instance, they are key-aspects of robots that are required to interact with human workers in not confined environments. The introduction of a mechanism that passively compensates the joint torque caused by the weight of the robot may offer a valid solution. Relieving the actuators from generating the torques due to gravity has the consequence of decreasing the power consumption and allowing the possibility to reduce the size and weight of the actuators. Thus, an overall lighter robot is obtained, further reducing the power consumption. Furthermore, in the context of human-robot interaction a lighter robot is potentially less dangerous and can be manoeuvred at higher speeds. In addition, a gravity compensator allows the robot to hold a static position without the need of an external power source, hence avoiding the risk of collapsing in case of failure of the actuators. This work is focused on a gravity compensator composed by a noncircular pulley and a spring. This specific gravity compensator features the possibility of generating an arbitrary torque profile, so it is not limited to compensate exclusively the trigonometric torque profiles due to gravity on revolute joints. First, a virtual prototype of the mechanism applied to an inverted pendulum was developed. This model was initially used to study the static and dynamic behaviour of the mechanism in order to validate its theoretical concept. Following, it was used to guide the design of a physical prototype of the system where the inverted pendulum was actuated by an electric motor. In designing the test bench, particular attention was made to include a system to effectively measure the pretension of the springs. Once that the test bench had been assembled, a set of experiments was performed to verify the results obtained from the simulations of the virtual model. Finally, the results are discussed along with problems, limitations and possible improvements of both the virtual model and the physical prototype.