Doktorsavhandling, 2020

This thesis highlights our efforts in energy and route optimization of moving devices. We have focused on three categories of such devices; industrial robots in a multi-robot environment, generic vehicles in a vehicle routing problem (VRP) context, automated

guided vehicles (AGVs) in a large-scale flexible manufacturing system (FMS).

In the first category, the aim is to develop a non-intrusive energy optimization technique, based on a given set of paths and sequences of operations, such that the original cycle time is not exceeded. We develop an optimization procedure based on a mathematical programming model that aims to minimize the energy consumption and peak power. Our technique has several advantages. It is non-intrusive, i.e. it requires limited changes in the robot program and can be implemented easily. Moreover,

it is model-free, in the sense that no particular, and perhaps secret, parameter or dynamic model is required. Furthermore, the optimization can be done offline, within seconds using a generic solver. Through careful experiments, we have shown that it is possible to reduce energy and peak-power up to about 30% and 50% respectively.

The second category of moving devices comprises of generic vehicles in a VRP context. We have developed a hybrid optimization approach that integrates a distributed algorithm based on a gossip protocol with a column generation (CG) algorithm, which manages to solve the tested problems faster than the CG algorithm alone. The algorithm is developed for a VRP variation including time windows (VRPTW), which is meant to model the task of scheduling and routing of caregivers in the context of home healthcare routing and scheduling problems (HHRSPs). Moreover,

the developed algorithm can easily be parallelized to further increase its efficiency.

The last category deals with AGVs. The choice of AGVs was not arbitrary; by design, we decided to transfer our knowledge of energy optimization and routing algorithms to a class of moving devices in which both techniques are of interest. Initially, we improve an existing method of conflict-free AGV scheduling and routing, such that the new algorithm can manage larger problems. A heuristic version of the algorithm manages to solve the problem instances in a reasonable amount of time. Later, we develop strategies to reduce the energy consumption. The study is carried out using an AGV system installed at Volvo Cars. The results are promising; (1)

the algorithm reduces performance measures such as makespan up to 50%, while reducing the total travelled distance of the vehicles about 14%, leading to an energy saving of roughly 14%, compared to the results obtained from the original traffic controller. (2) It is possible to reduce the cruise velocities such that more energy is saved, up to 20%, while the new makespan remains better than the original one.

guided vehicles (AGVs) in a large-scale flexible manufacturing system (FMS).

In the first category, the aim is to develop a non-intrusive energy optimization technique, based on a given set of paths and sequences of operations, such that the original cycle time is not exceeded. We develop an optimization procedure based on a mathematical programming model that aims to minimize the energy consumption and peak power. Our technique has several advantages. It is non-intrusive, i.e. it requires limited changes in the robot program and can be implemented easily. Moreover,

it is model-free, in the sense that no particular, and perhaps secret, parameter or dynamic model is required. Furthermore, the optimization can be done offline, within seconds using a generic solver. Through careful experiments, we have shown that it is possible to reduce energy and peak-power up to about 30% and 50% respectively.

The second category of moving devices comprises of generic vehicles in a VRP context. We have developed a hybrid optimization approach that integrates a distributed algorithm based on a gossip protocol with a column generation (CG) algorithm, which manages to solve the tested problems faster than the CG algorithm alone. The algorithm is developed for a VRP variation including time windows (VRPTW), which is meant to model the task of scheduling and routing of caregivers in the context of home healthcare routing and scheduling problems (HHRSPs). Moreover,

the developed algorithm can easily be parallelized to further increase its efficiency.

The last category deals with AGVs. The choice of AGVs was not arbitrary; by design, we decided to transfer our knowledge of energy optimization and routing algorithms to a class of moving devices in which both techniques are of interest. Initially, we improve an existing method of conflict-free AGV scheduling and routing, such that the new algorithm can manage larger problems. A heuristic version of the algorithm manages to solve the problem instances in a reasonable amount of time. Later, we develop strategies to reduce the energy consumption. The study is carried out using an AGV system installed at Volvo Cars. The results are promising; (1)

the algorithm reduces performance measures such as makespan up to 50%, while reducing the total travelled distance of the vehicles about 14%, leading to an energy saving of roughly 14%, compared to the results obtained from the original traffic controller. (2) It is possible to reduce the cruise velocities such that more energy is saved, up to 20%, while the new makespan remains better than the original one.

Energy optimization

industrial robots

vehicle routing problems

automated guided vehicles.

scheduling

routing

online participation

Opponent: Professor Maria-Pia Fanti, Polytechnic of Bari, Italy

Chalmers, Elektroteknik, System- och reglerteknik, Automation

11th IEEE International Conference on Automation Science and Engineering, CASE 2015, Gothenburg, Sweden, 24-28 August 2015,; (2015)p. 1345-1350

**Paper i proceeding**

IEEE Transactions on Automation Science and Engineering,; Vol. 14(2017)p. 646-657

**Artikel i vetenskaplig tidskrift**

IEEE Transactions on Automation Science and Engineering,; Vol. 16(2019)p. 127-137

**Artikel i vetenskaplig tidskrift**

IEEE International Conference on Automation Science and Engineering,; Vol. 2018-August(2018)p. 92-97

**Paper i proceeding**

IEEE International Conference on Automation Science and Engineering,; Vol. 2019-August(2019)p. 891-896

**Paper i proceeding**

IEEE Transactions on Automation Science and Engineering,; (2020)

**Artikel i vetenskaplig tidskrift**

We have focused on three categories of such devices; industrial robots, generic vehicles, and

guided vehicles (AGVs) in a large-scale flexible manufacturing system (FMS).

In the first category of moving devices, that is, industrial robots, we have tackled the challenge of reduce energy consumption of robots in manufacturing industries with promising results. The results show that our optimization reduces up to 30% of energy consumption and up to 60% in

peak-power. We show that it is possible to decrease energy consumption and peak-power without changing neither cycle time nor path. Our technique comprises logging robot motions, optimizing it, and then running the new trajectory. What distinguishes our work from similar attempts is that it does not require system identification or confidential robot data. We have tested our method on several different types of robots in our lab at Chalmers, and in labs at the German companies KUKA Robotics GmbH, and Daimler AG.

Regarding routing problem of generic vesicles we have two contributions. The first one is that we have combined known optimization algorithms with another one to improve the performance. The results show that the combination works well. Moreover, we have parallelized the developed algorithm to benefit from the existing multi-core processors, to speed up the calculations.

Finally, in the third category of moving devices, we have dealt with AGVs. It is commonly believed that reduction of speed in AGV systems leads to lower system efficiency. However, it has been shown that speed management is an effective strategy to reduce energy consumption of mobile robots and robot stations. If one seeks to utilize the existing slacks in the schedule of the AGVs, it should be possible to reduce energy consumption without affecting system efficiency. It can furthermore be combined with better scheduling to even improve performance measures such as cycle time while reducing energy. In the thesis, we propose an optimization method that seeks to minimize a number of performance indexes for a real AGV system designed by a Swedish AGV manufacturer, which operates at Volvo Cars, Gothenburg. We also show that the optimization method allows for reduction of cruise speed while the mentioned performance measures are still better than the one obtained from the original traffic controller.

The thesis thus studies techniques for energy and route optimization of moving devices.

Datorteknik

Datavetenskap (datalogi)

Datorsystem

978-91-7905-286-7

Doktorsavhandlingar vid Chalmers tekniska högskola. Ny serie: 4753

Chalmers tekniska högskola