Energy-effective Programming of Collaborative Robots

The project has presented an EC model that estimates the electric power consumption using motor currents, motor speeds, and articulation positions. Knowing the robot's power profile, we find the EC by the integration of the power profile. The model performed accurately with a maximum RMS error of 6 [W] - 3.85%. One reason for the model error is the sensors' noise with a 1.7 [W] standard deviation. The model is reliable and can be used to estimate the energy consumption of instructions of a robot program.

The findings from AR exploratory study suggest that robot programming cannot be done using an AR application anytime soon. While some of the features were very helpful to the programmers, they are not useful for programming the robot. AR could help alongside the teach pendant to do the following: sales and marketing, commissioning, and troubleshooting.

Based on the exploratory study, an app for data collection, visualization and simulation was created. The data visualization and collection tools help to solve troubleshooting. If the robot data is stored and overlapped with the real robot, the operator has additional information to understand possible failures. Moreover, the AR digital twin of the robot help to determine the solution feasibility. Compared to current AR digital twins, our solution has the advantage of offering prediction of motor currents and energy consumption.

We conclude from the first method that it is recommendable to use linear movements in joint space (PtP) instead of cartesian linear movements, especially for fast moves. When the robot uses cartesian linear movements, the friction increases due to unnecessary movements. In our case of study, this method saves up to 37% (without payload) and 26% (with payload) of EC.

We proposed to obtain the characteristic curve (EC vs. execution time) of a given path. Then, it is possible to get the optimal scale factor that minimizes the robot EC. The results show that the robot consumes 96.71% (without payload) and 97.33% (with payload) of the maximum experimental energy consumption.

• The proposed control scheme reduces the energy consumption by 5.27% in the proposed experiment.

The dynamic power saturator transforms all the regenerated energy (potential energy or kinematic breaking energy) into kinematic energy. The proposed control scheme reduces the energy consumption by 5.27% in the proposed experiment. Besides, the temperature of the cabinet is reduced. Potentially, the energy for ventilation might be reduced.

We observed that the cobot joint configuration affects the EC. The experiments showed a correlation between the EC and the gravitational toque. The robot has a joint configuration that consumes less energy in the case of redundant positions. Additionally, for long waiting idles times, the robot should move to a comfortable position of minimal gravitational torque. This technique can reduce up to 18 % of the idle energy.

In this work, we analyze the reduction of energy consumption by improving the behavior of the components without structural changes by using cost-optimal reachability analysis. Due to the dependence of system behavior on the environment, we examine the behavior of the environment from two aspects. First, the behavior of the environment may be quite irregular, so using fixed, predefined probabilities for incidents makes the existing system much simplified. For this purpose, to increase the system’s flexibility against the disorderliness behavior of the environment, it is modeled dynamically. Second, the system can optimize its behavior by predicting the behavior that the environment may have in the future. The optimization performance was tested and verified using UPAAL-SMC, and the results showed that the energy consumption is reduced in 5 %. So, by examining the past environmental behavior, the system can dynamically adjust its behavior to optimize it in terms of energy consumption.