ORCAS: Optimized Robots Configuration and Scheduling System Marin Lujak CETINIA, University Rey Juan Carlos Madrid, Spain marin.lujak@urjc.es Alberto Fernández CETINIA, University Rey Juan Carlos Madrid, Spain alberto.fernandez@urjc.es ABSTRACT In this paper, we study a distributed intelligent multi-robot system (MRS) in assembly setting where robots have par- tially overlapping capabilities. We treat the problem of the system’s self-(re)configurability and self-optimization. In this light, we propose a distributed and optimized robots configuration and scheduling system ORCAS which inte- grates the MRS configuration based on semantic descrip- tions with process scheduling. Categories and Subject Descriptors I.2.9 [Robotics]: Workcell organization and planning; I.2.11 [Distributed Artificial Intelligence]: Multiagent systems Keywords Flexible robot systems; ontologies; semantic web; industrial robots; assembly; multi-agent systems. 1. INTRODUCTION In this paper, we study distributed and intelligent MRSs in assembly where robots have partially overlapping capabil- ities. In more detail, we tackle the issues of self-configuration and self-optimization and propose a distributed model which integrates the MRS configuration based on distributed se- mantic descriptions with distributed assembly process schedul- ing. The aim of the semantic configuration is to find feasible robots’ configurations which can satisfy customer demand. The assembly scheduling, on the other hand, determines robot-task assignments and sequencing of tasks assigned to each robot based on the minimization of the total assem- bly cost and time while respecting inventory constraints, task interrelations and the robot assembly capacities. The objective is to seamlessly optimize robots’ performance by dynamic reconfiguration and rescheduling thus minimizing assembly costs and off-line times due to changes in resource availability and product orders. Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for components of this work owned by others than ACM must be honored. Abstracting with credit is permitted. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. Request permissions from Permissions@acm.org. SAC’15 April 13-17, 2015, Salamanca, Spain. Copyright 2015 ACM 978-1-4503-3196-8/15/04...$15.00. http://dx.doi.org/10.1145/2695664.2696024 To increase distributed MRS’s autonomy and decrease the dependency on human operators, a possible way is to aid as- sembly process planning with knowledge representation and reasoning [5]. One of the recent achievements in robotics is a more intelligent execution of complex tasks by controllers with a high degree of autonomy through the network access to a knowledge repository and management system, e.g. [3, 5]. The scheduling problem is typically NP-hard and the scheduling of a robotic cell represents one of the most diffi- cult scheduling problems. A review of the state-of-the-art of recent research on dynamic scheduling can be found in [4]. To the best of our knowledge, the State-of-the-Art ap- proaches that integrate knowledge reasoning with assembly planning and scheduling are centralized methods that use centralized knowledge bases. In this paper, on the other hand, we propose a distributed semantic based approach. 2. PROBLEM DEFINITION We consider a MRS made of industrial robot manipulators with partially overlapping assembly capabilities producing a set of different types of products. Moreover, every product is semantically described by the user as a set of tasks to be executed on specific parts. Other than industrial robots, other types of devices such as conveyor belts, computer and numerically controlled machines and other stand-alone sys- tems such as inspection machines may also exist on the shop floor. Let a robot configuration be a combination of a set of compatible resources, i.e., robot arm, gripper, camera, and auxiliary material. Furthermore, let us assume there is a set of tasks that denote production processes (e.g., glue, attach, transport) of a particular type of a product and all the tasks are assumed different from one another. Tasks are made of more atomic tasks (actions) (e.g., for glue task: open gripper, use camera, insert glue material, align and push parts together). Furthermore, each task re- quires a specific robot configuration for processing depend- ing on the geometric, mechanical and physical parameters of the manipulated parts. This is why multiple resources within each robot configuration need to be combined se- mantically to perform a single task. We assume that the following information is given: as- sembly workflow and ontology, robots’ and tools’ ontology, and constraints on coupling of resources (tools and robots) in the robot configurations. Furthermore, we assume prede- fined user inputs for every assembly product and for every task for each time interval, holding, backlog, production, and other relevant costs. Given a finite time horizon of T time periods and assum-