Modeling and optimization of implementation aspects in industrial robot coordination
Introduction
The assembly line in the automotive industry usually consists of several automated cells where robots perform sealing and welding operations on the Body in White (BiW). Sensors, robots, and other equipment are usually coupled to a Programmable Logic Controller (PLC) that decides the timing of operations based on its programmed logic behavior. Besides moving the BiW, clamping, and controlling safety operations, PLCs are often used to schedule robot programs such that the robots do not collide with each other, see Fig. 1. This is done by signaling each robot whether it can continue its program or not, based on their occupancy of shared workspace volumes. The main rule is that only one robot at a time can occupy a common workspace, preventing any potential collision with other robots. This is a very critical task since the line throughput is heavily influenced by that and since incorrect behavior may result in robot damages, line stops, with all the corresponding drawbacks.
Therefore, an optimized schedule of robot operations is crucial and off-line simulation of robot programs plays a central role to achieve that. Indeed, simulation tools modeling the reality in a virtual environment are becoming a key factor for success in the very tough automotive market. Besides modeling and simulation, the most important feature in a simulation tool is the possibility to automatically or semi-automatically optimize the process.
We focus on this optimization aspect, applied to the minimization of cycle time in industrial robot stations, especially automotive, see [1], [2], [3]. In previous works, robot coordination has been achieved by changing the start time of each operation or robot motion, see [4], [5]: the nominal duration of each motion has been considered independent of its time scheduling. After this has been done in the virtual environment, the simulation engineer needs to transfer the obtained solution into the real world.
However, in reality, different hardware setups are adopted: robots communicate directly with each other through the robot controllers digital input/output, or they communicate through a PLC. Moreover, robot manufacturers might implement in different ways the instructions for signal exchanging. At the same strength, automotive industries can program differently how the handshaking between PLC and robot controllers behave. It turns out that these details have a large influence on the real cycle time: an aspect that has not been modeled and considered during the optimization, so far. Indeed, the robot might stop its motion to perform the handshake, increasing in a non negligible way the overall pre-computed cycle time. This communication aspect is becoming much more evident in Human-Robot applications where common tasks can be performed and must be coordinated in a safe way, see [6].
To resolve this neglected problem, we model the time delays related to the hardware specifications and consider them in a post processing step to minimize their influence on the cycle time.
This article is organized in the following way: first, a description and a mathematical model is given for the path coordination problem, in Section 2. Then the main problem is described in Section 3 where we highlight the importance to minimize the number of synchronization via points. Our contributions are presented in Sections 3.1 and 3.2, with different mathematical programming formulations, and in Section 4, with a new algorithm aiming to find near optimal solutions. In the last section computational experience and results are presented, based on two industrial scenarios. Finally, a conclusion section summarizes the results with some suggestions about future work.
Section snippets
Industrial robot coordination
The problem of finding a sequence of robot paths among several tasks in order to avoid collisions is quite general. In fact several problems need to be solved such as balancing the tasks among the robots, tasks sequencing, motion planning and robot coordination, see [1], [3], [7], [8], [9]. In addition to that, robot dress packs, that are a major reason for on-line adjustments, usually make the planning problem even more difficult, see [10]. Due to complexity of the problem, heuristic
Minimization of synchronization points
We have, so far, treated the modification of robot programs to coordinate them, assuming that the added instructions have no influence on the nominal optimal cycle time. Unfortunately, in practice, we have to deal with the implementation of such systems: the robots are usually connected to a PLC and, when they communicate with it, time delays can occur. These delays cannot be disregarded in bottleneck stations: they might be critical for the line throughput, having a direct impact on profits.
A
Improvement heuristic algorithm
Our major contribution is an algorithm to compute high quality solutions for large instances appearing in the studied applications, solving the problem modeled as (3). It is inspired by [25] and generalized to quadratic constraints. In fact the heuristic in Lan et al. [25] is designed for the SCP and cannot directly handle problems with quadratic constraints as in (3a). The removeAndUpdate function below shows how this can be done: the procedure holds even when each term in the sum is the
Results and discussion
In this section we report some of the computational experience for the method and algorithms presented in the previous sections. We have implemented the algorithms in C++ and interfaced with an offline robot simulation software. This platform offers the possibility to test real world scenarios, providing functionalities like distance queries, robot kinematic simulation and process optimization. Regarding the solution of the linear programming models we have used the Coin-Or package, see [26].
We
Conclusions and future work
In the area of commissioning robot programs for multi robot assembly and inspection, we presented solutions to problems arising when modeling robot communications that avoid inter-robot collisions. Two setup scenarios have been investigated: For the first one we provided an ILP model minimizing the number of wait points; for the second, quadratic constraints are added to the model and we minimize the number of via points where signal exchange happens. In order to solve the second problem we
CRediT authorship contribution statement
Domenico Spensieri: Conceptualization, Methodology, Software, Data curation, Writing - original draft, Writing - review & editing. Edvin Åblad: Methodology, Software, Writing - review & editing. Robert Bohlin: Conceptualization, Methodology, Writing - review & editing. Johan S. Carlson: Conceptualization, Methodology, Writing - review & editing, Supervision, Project administration, Funding acquisition. Rikard Söderberg: Supervision, Project administration, Funding acquisition.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgments
This work was carried out within the project Smart Assembly 4.0, supported by the Swedish Foundation for Strategic Research (SSF). It is also part of the Sustainable Production Initiative and the Production Area of Advance at Chalmers University of Technology.
References (32)
- et al.
Multi-robot spot welding cells for car-body assembly: design and motion planning
Robot. Comput. Integr. Manuf.
(2017) - et al.
Balancing a robotic spot welding manufacturing line: an industrial case study
Eur. J. Oper. Res.
(2017) - et al.
Productivity/energy optimisation of trajectories and coordination for cyclic multi-robot systems
Robot. Comput. Integr. Manuf.
(2018) - et al.
Path planning using lazy PRM
IEEE International Conference on Robotics and Automation
(2000) - et al.
Principles of Model Checking
(2008) - et al.
Load balancing of welds in multi station sheet metal assembly lines
Proceedings of the ASME 2010 International Mechanical Engineering Congress & Exposition, Vancouver, British Columbia, Canada
(2010) - et al.
Coordination of robot paths for cycle time minimization
Proceedings of the IEEE International Conference on Automation Science and Engineering, Madison, USA
(2013) - et al.
A software tool for human-robot shared-workspace collaboration with task precedence constraints
Robotics and Computer-Integrated Manufacturing
(2021) - et al.
Route Planning for Robot Systems
Operations Research Proceedings
(2011) - et al.
Solving a vehicle routing problem with resource conflicts and Makespan objective with an application in car body manufacturing
Optim. Methods Softw.
(2014)
Task assignment, sequencing and path-planning in robotic welding cells
Proceedings 18th International Conference on Methods and Models in Automation and Robototics
Quasi-static path optimization for industrial robots with dress packs
Robotics and Computer-Integrated Manufacturing
Robot coordination using task-priority and sliding-mode techniques
Robotics and Computer-Integrated Manufacturing
An iterative approach for collision free routing and scheduling in multirobot stations
IEEE Trans. Autom. Sci. Eng.
Robot Motion Planning
Planning Algorithms
Cited by (9)
Recovering feasibility in real-time conflict-free vehicle routing
2023, Computers and Industrial EngineeringA Three-Stage model for Clustering, Storage, and joint online order batching and picker routing Problems: Heuristic algorithms
2023, Computers and Industrial EngineeringNetwork-based dynamic dispatching rule generation mechanism for real-time production scheduling problems with dynamic job arrivals
2022, Robotics and Computer-Integrated ManufacturingCitation Excerpt :Since this method makes scheduling decisions only upon the current system state information without prescheduling schemes, it has good practicability for the actual production environment with frequent dynamic events and high uncertainty. Heuristic rules have therefore become the primary solution to real-time production scheduling problems because of their advantages such as simplicity, efficiency and quick response [25]. However, the current heuristic rule design method is mainly based on manual experience, and knowledge related to complex production scheduling problems is difficult to obtain, making it extremely difficult to develop efficient scheduling rules [26].
A workload identification method of industrial robot combining dynamic model and convolutional neural network
2024, Engineering Research Express