Admittance Controller on Hapitc Paddle

In this project, an admittance controller was implemented, a system linking input force to motion. Unlike the common impedance controller used by other groups, this approach suits scenarios where plant dynamics are less known or challenging to model accurately.

The key advantage of the admittance controller lies in its ability to achieve high impedances with low control gains, overcoming limitations of the impedance controller. The controller interprets user-applied force as an input to a virtual model, generating a simulated motion represented by 𝑚 ⋅ 𝑎 + 𝑏 ⋅ 𝑣 + 𝑘 ⋅ 𝑥 = 𝐹. This virtual motion is seamlessly integrated into the Haptic Paddle's control scheme using velocity control (PID). The end-effector experiences a blend of physical and virtual dynamics. The goal is to identify stable human-robot interaction regions, represented in an M-B-K plot. Equipped with a force sensor, the Haptic Paddle measures the necessary human-robot interaction force for the admittance controller. During the presentation, the admittance controller's concept, implementation, and performance will be discussed, comparing it to the simpler impedance controller.

Within the scope of the project, a virtual wall was dynamically generated with a 15-degree inclination. The validation process involved experimenting with different mass values. The video demonstrations vividly showcase the expected trends: lower mass values, such as 0.1 kg, induce unstable behavior in the virtual wall, while a progressive increase in mass imparts greater stability.

A noteworthy observation from the simulations is the inverse relationship between mass and recovery time. Specifically, as the mass value increases, the haptic paddle takes a more deliberate pace to recover from exceeding the boundaries of the virtual wall. This phenomenon is attributed to the higher integration time associated with higher gains. The intricacies of these interactions are thoughtfully captured and presented in the accompanying video documentation.

Admittance controller with a virtual wall at 15°:

M = 0.1 kg (instability):

M = 1 kg:

M = 100 kg:

M = 1000 kg (very stable performance):