Seamless human robot collaborative assembly – An automotive case study

Highlights

• Implementation of a robotic system for advanced human robot collaborative assembly.

• Use of wearable devices for multi – modal interaction and human robot communication.

• Deployment of a safety concept involving multiple robot safety control functions.

• Deployment of novel integration and orchestration mechanisms based on SoA concept.

• The complete system has been validated in a case study from the automotive industry.

Abstract

This paper presents the implementation of a robotic system for advanced human robot collaboration assembly and discusses all the technological approach that has been implemented for facilitating the interaction and support of human operators. Unlike current industrial practice where the assembly is performed by operators, the proposed approach aims at combining the benefits of high payload industrial robots with human capabilities under a fenceless environment, by assigning to them each task based on their capabilities. Enabling technologies involve manual guidance techniques and new wearables devices allowing for multi – modal interaction as well as robot safety control functionalities. Wearable devices such as Augmented Reality glasses and smartwatches are used for closing the communication loop between operators and robots under a service-oriented architecture. The complete system is validated in a case study from the automotive industry under the ROBO-PARTNER project. A detailed safety analysis of the scenario has been performed supported by a Risk Assessment, safety concept and Safety Related Parts/Control Systems (SRC/CS) design. The findings support the concept that humans’ and robots’ destiny is collaboration rather than competition.

Introduction

In an attempt to follow the requirement for more customized products [1] and smaller lot sizes [2], production and research engineers have turned to the concept of Human Robot Collaboration (HRC) [3]. The benefits of implementing such production cells lays in the implementation of flexible and highly reconfigurable production systems [4] which can easily change their operation to accommodate different product families, similar to the way that a human operator would do. In this direction, projects such as the EC funded ROBO-PARTNER [5] have focused on integrating new forms of interaction between robots and workers aiming to make the most out of the synergy effect. This means to efficiently combine and exploit the robot's precision, repeatability and strength with the human's intelligence and flexibility [6].

Such attributes are indispensable when considering the sustainability of small and medium sized (SMEs) production firms who have to bear the costs for re-equipping or modifying their current production systems [7].

In order for human operators to collaborate with the robot, the means to safely interact with it need to become available. Human Robot Interaction can be perceived as the exchange of information between the human and robot and can be implemented using different concepts [8]. A high-level classification of the interaction modes may involve the following:

• Direct physical interaction with the robot or the part being processed by both human and robot: This includes the case where the human's body and/or hands are in contact with the robot or the part and the two actors can coordinate their motions to achieve the task. A typical example involves the case of manual guidance where the robot is carrying a large part and the human can accurately guide the robot to the assembly/process position by applying small forces on the robot or the part itself [9], [10], [11], [12], [13], [14]. Another example is the case of impedance control, where small forces applied to the part during a screwing operation are compromised by the robot motors [15].

• Remote contactless interaction: This category involves the interaction through interfaces, such as voice or gestures recognition software as well as 3D cameras, being able to translate human input into actions for the robot. For instance, a voice command for the robot to move its end effector up, would have to pass through a recognition, interpretation, dispatching, motion planning and execution process in order to be actually performed [16], [17], [18].

• Teleoperation: Teleoperation differs from the previous category in the sense that the operator is directly driving the robot through an interface (joystick, teach pendant etc) determining its position and velocity at all times. There is no need for intermediate software/controller to translate the human input [19], [20].

• Message/ information exchange through human machine interfaces or other IT systems: This category is closer to the current practice where the robot exchanges information with the use of digital I/O signals, transmitted through a PLC or physical buttons located in the cell. Recently, the trend is to enable more content rich messages through services in order to have more flexibility in the design of the robot side interpretation and programming modules [21], [22].

By employing the aforementioned interaction methods and introducing means for coordinating the activities of humans and robots, a collaborative task execution can be achieved. HRC has been implemented in the past under different paradigms and concepts depending on the task to be achieved as well as the different types of environment layout and the desired collaboration to be implemented. Under this context, hybrid production/assembly operations are classified into: shared tasks and workspace, common task and workspace and common tasks and separate workspace [23].

In many low-volume production settings, direct physical interaction (or human-guided assembly) has significant advantages but human cognitive capabilities must be sensibly joined with the robot system's force and precision [24], [25]. Such hybrid assembly systems can be divided into workplace sharing systems and workplace and time-sharing systems. Other classification involves robot assistants as a directly interacting partner [26] collaborative robots (co-bots) (mechanical devices that provide guidance using servomotors, while a human operator provides motive power) [27] and portable robots where the workspace is not restricted to the place of its actual use [28]. Further research has investigated power amplifying assistance devices as well as various two-arm configurations [29], [30] including the famous Baxter robot unveiled in late 2012, that uses tactile feedback.

The enablers of the HRC paradigm have been (and are being) investigated by recent research projects (ROBO-PARTNER [5], [31] LIAA [32], THOMAS [33], SYMBIOTIC [34] etc.) while several new robotic products have been introduced to the market aspiring to cover the need for collaborative applications. The focus has been mainly given to:

• 1. lightweight robotic arms such as the UR10 [35], KUKA IIWA [36], SAWYER [37], DLR [38] etc.

• 2. sensing equipment involving force/torque sensors, cameras and 3D depth sensors for perception, capacitive sensors, infrared/proximity detection devices etc.

• 3. safety equipment for the preservation of human operators’ health including laser scanners, safety curtains/light barriers, pressure sensitive skins etc. [39].

However, bringing together humans and robots is not as trivial as it may sound. Robots are machines that operate in a relatively large workspace and depending on their configuration and tooling, hazardous situations may arise. Even in the case where robotic arms operate in a completely pre-defined way, the unpredictability of human behavior may still result in dangerous interaction. Thus, there is a great need of structuring collaboration schemes that coordinate the operation between these two production entities. A first approach to establish collaboration schemes that can guarantee the safety of human operators has been captured in standards that aim to regulate the interaction between humans and automation.

Originally, ISO 10218 Part 1 (“Safety of Robots”) and Part 2 (“Safety of Robot Integration”) were intended to address workplace safety requirements for “assisting” robots working in a “collaborative workspace” with users. The two parts describe basic hazards associated with robots and provide requirements to eliminate, or adequately reduce, the associated risks. In this standard, the following requirements are included: a) Safety – related control system performance, b) robot stopping functions, c) speed control, d) operational modes, e) collaborative operation requirements and f) axis limiting. The second part foresees further requirements on robotic integration such as: a) general, b) limiting robot motion, c) integrated manufacturing system interface and d) collaborative robot operation [40], [41].

The most recent ISO/TS 15066 (Robots and robotic Devices – Collaborative Robots) provides more concrete guidelines for collaborative robot operation in shared workspaces with humans: a) establishing minimum separation distance, b) establishing maximum safe speed, c) tracking operator position and velocity, d) determining and avoiding potential contact, e) avoiding potential collision, f) operator controls, g) power and force limiting, h) technological, medical/biomechanical requirements and ergonomic requirements and so forth [42].

These days a variety of manufacturing industries aspire to introduce it in their production lines using the standards mentioned above as guidelines. BMW automotive industry, introduced in their manufacturing plant cooperative robots to work around human operators in order to take over tasks that could cause human workers repetitive strain injury [43], [44], [45], [46]. Universal Robots has been applied in BMW premises without fences in vicinity of a human worker in order to perform the door sealant. The same OEM aims to introduce collaborative robots to serve human workers as assistants handling their tools and parts during assembly [47]. Also, VW has integrated an UR – 5 in the cylinder head assembly section of the plant for inserting spark plugs into the cylinder heads. Additionally, Audi AG also, in order to optimize ergonomical issues and automate routine operations, has introduced a collaborative KUKA robot internally named “PART4you” to work in close cooperation with the human operators providing them important assistance in assembly operations [48].

When reviewing the latest applications in HRC, it can be observed that a lot of the safety barriers have been lifted as system integrators have found ways to meet the requirements set by the aforementioned standards. However, there are some very important gaps identified:

• In the majority of HRC applications lightweight robots are used. These robots are indeed easier and safer to work with as they inherently are not able to exert large forces and they incorporate direct drive motors with embedded force torque sensors that allow much higher sensitivity in detecting collision. This however trims the original vision of exploiting robot capabilities and especially the fact that they are high powered machines that can undertake strenuous tasks (lifting, prolonged holding, working in hard to reach areas etc.). Industrial robots on the other hand are perfectly capable of offering strength augmentation and task support [49] but are constrained to work behind fences as there are no tools available to guarantee safety. The ergonomy benefits that arise from such applications can be significant.

• The implementation of safety functionalities should not obstruct the workflow. The design of protective function should be such that would minimize the need for recovery even when human error is present. For instance, in the case of manual guidance by the operator, the robot controller should ensure that the motion is constrained and no impact with surrounding objects is possible rather than engaging Safety Stop conditions and requiring a manual reset of the system to a “safe condition” where the operation may restart. The operator should also be allowed to intervene in any stage of the workflow without stopping the execution of the task for long periods of time (e.g. to manually drive the robot to the next start positions etc.)

• Approaches claiming to eliminate physical barriers by replacing them with laser scans or light barriers are not sufficient as they still forbid the human from accessing areas inside the robot workspace. A more dynamic monitoring approach is needed to ensure that the workspace is adjusted according to the actual status of the robot and the task that it is performing.

• Another significant gap in the current practise is the lack of means to efficiently integrate human operators in the collaboration workflow with the robot. Currently the interaction with the robot inside the shopfloor is achieved through warning lights and human machine interfaces, presenting the current status of the robot usually in a format that is not designed for non-expert operators (e.g. robot code). On the other hand, human feedback is implemented in the form of push buttons located away from the robot in order to ensure safety. As a direct result, the operator is not allowed to focus on the task at hand and spends a significant amount of time trying to remember or predict what the robot will do next or move back and forth to the assembly area.

It is evident that in order to establish a seamless collaboration, viable from a sustainability and performance point of view, several technologies need to be brought together.

Focusing on medium and high payload industrial robots, the ROBO-PARTNER project set out in 2013 to investigate the feasibility of bringing together these powerful machines and the human operators. This paper presents the approach of the project in integrating different HRI and safety technologies to implement HRC concepts that can be combined to fully implement assembly scenarios. Driven by actual end user needs for handling and assembling heavy parts (i.e. vehicle axle and wheel groups) with parts requiring dexterity (screws, cables etc), the project provided both hardware and software solutions to perform the actual assembly as well as the monitoring and coordination of the activities in the cell. The main targets set were:

• To reduce the ergonomic impact, both physical and cognitive, of assembly operations that are now carried out only by operators

• To enable the safe collaboration between humans and industrial robots, with the latter acting as assistants, undertaking demanding tasks. This is achieved by both new safety monitoring concepts as well as the introduction of a service-based control architecture that allows bidirectional communication through multiple user-friendly interfaces.

• To provide the tools that allow the immersion of the human in the workflow and increase the awareness of the robot's operation status and underlying safety functions. The aim is to increase the feeling of comfort and acceptance of the operators when working with such powerful equipment.

• To support the operators in handling multi variant production that is now easier to implement with the use of flexible resources such as the robots by providing him with the right amount and type of information using state of the art technologies.

The following sections are organized as follows: Section 2 provides the description of the approach that was used to analyses the requirements for safety and interaction towards implementing the hybrid assembly cell. In Section 3 the different technologies that have been developed and integrated within the actual cell are discussed. The execution of the assembly scenario with the use of the developed modules is outlined in Section 4 and the evaluation of the achieved results is presented in Section 5. The limitation of the approach, the industrial exploitation potential and the areas for future work are highlighted in Section 6.