Keywords

1 Introduction

The last decades brought new human-robot collaboration (HRC) areas in the same working space with examples in the industry, education, agriculture, healthcare services, security, and space exploration [1]. The collaborative robots (so-called cobots) are an important part of the 4th industrial revolution, better described as Industry 4.0 [7]. Unlike the well-known industrial robots, cobots are simple in operation, lightweight, reliable with enhanced sensors and easy to deploy. They are designed to follow special safety measures, handling human collision detection and avoidance contrary to closed industrial robotic cells.

Most cobots adopt compliant mechanisms and lightweight designs to reduce the impact force once collisions occur. Other solutions are power and force limiting (PFL) methods based on temporal dynamic models, and current signals from embedded torque sensors [8]. Compared with the above methods, equipping cobots with the electronic skin (e-skin) is a complementary way to increase their safety levels [4, 5].

The sense of touch in robotic devices has been developed for many years. Mostly as a part of a gripper or an end-effector mechanism. Adding touch allows bio-mimicry to robots, as it has a very important defence function in nature. In animals, when the harmful factor causing the pain is triggered, the body automatically moves to avoid contact with the pain-causing factor. This movement is unconditional (it occurs automatically and is not subject to will). Recently, with the sensor miniaturization and new tactile technologies development, there is a possibility for the full robot body tactile e-skin manufacturing, and usage [4].

The paper aims to describe the developed hardware and software framework, combined in a laboratory test stand, allowing the functional tests of the newly developed graphene-based e-skin solution with the cobot. During tests, the e-skin was separated physically from the robot.

The paper is organized as follows. In Sect. 2, the prototype consisting of cobot, e-skin and software integration is presented. In Sect. 3, the preliminary tests are described and the obtained results are given. Finally, Sect. 4 provides the summary and further investigation proposal.

2 Developed Test Framework

The hardware framework of the developed laboratory test stand consisted of the e-skin based on graphene nanoplatelets and an ES5 robot from EasyRobots companyFootnote 1. For tests, the e-skin module was placed outside the working space of the manipulator. It was connected to the robot via a desktop computer with an Ethernet interface. Data exchange between the computer and the e-skin was based on UART communication using the USB port. The software framework was developed using Robot Operating System (ROSFootnote 2) [2, 6].

2.1 E-skin

The e-skin was constructed of force-sensitive resistors (FSR) connected in a rectangular matrix registering a pressure map from the touch exerted on it.

The e-skin consisted of two layers, presented in Fig. 1 [4, 5]. The first one is a conductive layer of comb electrodes (1b-1) printed on plastic foil and connected along columns and rows. The second one consisted of FSR sensors arranged in a rectangular pattern (1b-2) placed on a plastic foil. In the research, we have used FSR matrices with the size of a single sensor approx. 5 \(\times \) 5 mm. In order to present the touch measurement, a matrix of FSR sensors with the dimensions of 16 \(\times \) 32 cells was used.

The FSR matrix combined the individual FSR sensors into columns and rows. In order to estimate the location of the pressure applied to the FSR matrix, each of the columns and rows was separately, sequentially powered by a dedicated electronic controller supplying \(V_{cc} = 3.3\) V to only one sensor. The output \(U_{x}\) (see Fig. 1a) of the matrix was connected to the ground via a measuring resistor. Accordingly, a circuit consisting of only one column and one row could be closed by the selected FSR sensor pressed to a row and column comb electrodes. Then it was possible to read the measuring signal \(U_{x}\) from the selected sensor. It has been estimated that as a result of the pressure, the measured FSR resistance changed in the range of \(300\, \Omega {-}3.5\, \mathrm{k}\Omega \). Each FSR was based on graphene nanoplatelets similar to that described in [3]. The FSR matrix was connected with comb electrodes to detect the place and the contact force exerted thereon.

Fig. 1.
figure 1

a) Measurement principle for FSR matrix of size 2 \(\times \) 2 [4]; b) the e-skin [5]: 1 - a conductive layer of comb electrodes printed on plastic foil, 2 - FSR sensors arranged in a rectangular pattern placed on a plastic foil.

Full size image

2.2 Collaborative Robot

We have tested the e-skin prototype using an ES5 robot. It is an industrial 6-axis robot with a ROS-based communication interface, mainly used in the metalworking industry, for palletizing, packaging and welding (Fig. 2c).

Fig. 2.
figure 2

The e-skin software (a), robot in simulation mode (b), a real ES5 industrial robot (c).

Full size image

2.3 Software Integration

In the research, we have used a data exchange protocol based on publishing and subscribing messages provided by ROS. As part of the software framework, the ROS node was developed, allowing the operation of the e-skin with the robot. The data package was cyclically sent (with a frequency 50 Hz) in the form of a two-dimensional matrix of numbers corresponding to the e-skin prototype surface fields, with values reflecting the pressure level. A separate ROS node associated with the robot control system was analyzing the received values seeking maximum values. If it exceeded a fixed threshold value, the node sent the information into a robot control system to slow down the robot movements.

3 Tests

3.1 Research Method

We carried out initial research using a robot simulation mode and a prototype of the e-skin module. We observed the interaction of the developed e-skin prototype (Fig. 2a) with the robot system using visualization in RViz (ROS Visualization, see Fig. 2b).

The second stage of the research was the integration of the device with the real robot. We have tested the correctness of the algorithm for forcing the robot to switch to slow speed mode in the effect of human contact with the surface of the e-skin and finding the robot’s reaction time to a change in the state of the signal transmitted from the e-skin moduleFootnote 3.

In both research stages, the robot moved along an experimental trajectory consisting of sections of straight lines run through the tool centre point (TCP) with a maximum speed of 1 m/s. Then, in response to human contact with the e-skin prototype matrix, the robot was forced to slow down to 250 mm/s (TCP speed, following the guidelines in point 5.7.3 of the PN-EN ISO 10218-1 standard, defining the collaborative robot reduced speed).

3.2 Research Results

Figure 3 shows the result of the robot reaction time to the pressure from the touch exerted on the e-skin module. It indicates the moments of pressure detection, receiving this information by the robot, and achieving the reduced speed. The individual times are shown in the Table 1. It can be seen that the total time from the detection of the force applied to the e-skin until the robot reaches the reduced speed is 0.414 s.

Fig. 3.
figure 3

Reaction times for pressure detection

Full size image
Table 1. Time reactions for e-skin pressure detection.
Full size table

4 Summary

The paper describes the hardware and software framework for the newly developed graphene-based e-skin connected with an ES5 robot manipulator. The e-skin was connected to the robot main computer unit within the Robot Operating System. The functional tests on a laboratory stand showed that the robot could be forced to slow down to 250 mm/s for the TCP speed in response to human contact with the e-skin. The presented tests prove that the developed system can be used for further research of the developed e-skin technology, as it also allows for new ways of interacting with electrical devices of daily use and can improve safety in traffic communication when introduced as an additional body coating.

The developed test framework will be the basis for further integration of the robot control system with the e-skin, which will allow for safe human-machine cooperation. A more elaborate robot response based on human touch is possible but requires improvements in the presented system’s performance. In further works, the analysis and optimization of measurement and computational delays should be considered. The robot’s reaction time is currently too low for human to work with it comfortably. In terms of the robot’s response to human touch, the reaction time should be shortened to millisecond values.

Another part of further research will also cover the integration of the e-skin module with the robot control system by covering the whole surface of the robot structure with the flexible e-skin shell. For this purpose, a flexible cover with the applied sensory system, the same as the prototype described in this work, will be printed on the elastic mat, and a unique wiring system will be developed.

Finally, in presented research, we do not model the current flowing through sensors. It was observed that if many e-skin sensors are pressed simultaneously, the largest current will flow through the selected sensor, and a smaller current will also flow through the other sensors. It can lead to the ambiguity of the readouts. However, tests demonstrating the operation of the e-skin module have shown that it is still possible to determine the location and estimate the value of the pressure applied to the selected FSR. Addressing this problem is planned as part of the further e-skin module development.