Sensory grippers: a handyman technology project.

Bilberg, Arne ; Jones, Richard William

Abstract: The HANDYMAN project is a Southern Denmark government sponsored initiative that aims to help improve the automation capability of small businesses. This contribution provides the background and motivation behind a project developed by the Mads Clausen Institute within HANDYMAN which is currently investigating the development of new technology to improve the sensing capability of robot grippers.

Key words: Automation, Robotics, Grippers, Sensor Technology.

1. INTRODUCTION

The HANDYMAN project is a Southern Denmark government sponsored initiative, involving a consortium of Engineering firms and University research institutes that aims to help improve the automation capability of small businesses. The ultimate aim is to improve the competitiveness of these businesses in the European marketplace.

Two types of projects currently exist within HANDYMAN--Application and Technology. Application projects are short-term projects wherein a business requests an improvement to a current process, e.g. automating a manual operation or modifying the current automation process. The consortium develops a prototype solution within 6 months. Current projects include automating the packing of flowerpots (with flowers) into trays and the picking up of randomly stored tubular components and placing them in a leak testing machine with the correct orientation. Technology projects are longer term, up to 3 years to develop a prototype solution, and are therefore more research orientated. The aim of these Technology projects is to examine future automation requirements and develop new technology.

This contribution provides an overview of a Technology project developed by the Mads Clausen Institute within HANDYMAN, namely improving the sensing capabilities of robot grippers.

2. MOTIVATION AND PROJECT DESCRIPTION

The grasping or picking up arbitrary objects with a robot gripper remains a difficult task with many open research problems. Throughout the 1980's and 1990's it was expected that tactile sensors--sensors that measure the given properties of an object through physical contact between the sensor and the object--would be developed, commercialised and integrated into automated manufacturing processes, via the robot gripper, in great numbers. This did not happen because in highly structured, predictable manufacturing environments the only sensors usually required are those used to detect error conditions (Lee & Nichols 1999). Recent developments in contact sensors for manufacturing industry concentrate solely on sensors that help report accurate position or force information of the gripper, see for example (Machine Design, 2007). Tactile sensing has still found a role in robotics when the environment is unstructured such as the natural world (agriculture, and food processing) and the loosely constrained environments found in human workplaces such as hospitals, companies and to a lesser extent the home. Some of the current Application Projects within HANDYMAN are concerned with developing automation in these unstructured environments and future similar applications such as the automated picking of apples are being considered.

To provide flexible automation solutions in these types of environments will require robotic handling systems with high sensitivity, adaptation and dexterity. To grasp, manipulate or process soft objects such as fruit, confectionary, and food items, robotic devices will have to sense hardness, textures, and surface properties.

This project aims to develop improved sensing technology for industrial robot grippers with the long-term aim being that grippers will eventually have the ability to sense the properties of the object and then make decisions about how to grip, or perhaps not to grip, the object. In addition, sensors that provide diagnostic information about a simple operation, e.g. placing a flowerpot in a tray, can tell us if the operation has been successfully achieved and if not why not?

Possible uses of this technology could be the gripping of sensitive objects without damaging the object such as fruit picking. In fact the sensory gripper could perhaps even make decisions about whether to pick or not to pick the fruit based on an assessment of its maturity. In the automated sorting of clean laundry the sensory gripper could make distinctions between the type of fabric in the washed article and sort the article appropriately.

3. STATE OF THE ART

Research into the development of tactile sensors that provide feedback information on the contact forces exerted upon an object and to detect and correspondingly react to the possible slip of a gripped item has been on-going for many years, see for example (Tise, 1988; Zhu & Spranck 1992). Companies such as The Shadow Robot Company, Precision Profile Systems and the Nitta Corporation have successfully commercialised tactile sensors so that they are now readily available. Some doubts still remain about the life expectancy of these devices though with many prototype devices having soft coverings that wear relatively easily.

If the object material is compliant when gripped then tactile sensing based-approaches can also provide information on the mechanical properties of the object material. Research in this area is more recent (Shikida et al., 2003; Salo et al., 2007). For example one of these sensors (Shikida et al., 2003) consists of a diaphragm, a piezo-resistive displacement sensor on the diaphragm and a chamber for pneumatic actuation. An array of these sensors, can in theory, detect the two-dimensional contact force distribution and hardness distribution and surface texture of the contacted object.

There has an enormous amount of research done, a large proportion of it being driven by the medical sector, on trying to mimic human haptic exploration. This is the mechanism by which we learn about the surface properties of unknown objects, see for example (Maekawa et al., 1995). A large number of recent tactile sensor developments aim to mimic the sensing ability of the fingertip wherein the total force exerted on the object as well as the local static strain between the sensor and the object surface can be detected (Tada et al., 2003; Schmidt et al., 2006). A sensor with these dual abilities is able to detect any slip on the object in the gripper fingers. Okamura and Cutosky (2001) have developed an approach for haptic exploration of unknown object surfaces with robotic fingers whereby they both explore and manipulate the object.

With regards to the gripper itself it has to be mechanically able to grip a wide range of objects via grabbing or picking. Add in the demands of haptic exploration of the object surface and this would imply the use of at least a three-fingered gripper The 3-fingered industrial BarrettHand (Machine Design, 2001) is currently the most prominently used gripper for intelligent gripping research.

The control of the gripper fingers is a major research area in itself, for example see (Tise, 1988). Haptic exploration that also carries out manipulation can be a very complex control problem though Okamura and Cutkosky (2001) simply alternate the manipulation phases with the exploration phases which simplifies the control problem. The dynamic control of a three-fingered robot gripper manipulating an object while allowing one of the three fingers to slide in order to change its grasp location on the objects surface is formulated in (Zheng et al., 2000).

This section has concentrated on contact sensing but within the project we also want to explore the possibility of integrating non-contact sensing with any developed tactile sensing solution.

4. THE RESEARCH CONTRIBUTION

The sensor development work is divided into two areas, contact and non-contact sensing. We are currently investigating the appropriateness of existing tactile sensing technology and identify its advantages and disadvantages for the possible range of applications we are interested in. Part of this is being done in conjunction with the application project on flowerpot packing where a range of tactile sensors have been applied to the robot gripper to a) gain experience with different tactile sensor technology and b) to develop a simple sensor-based diagnostic tool that indicates if the flowerpot packing operation has been successful. For non-contact sensing there are many possibilities, depending upon what non-contact information will be most useful for gripping (or non-gripping) purposes. Currently we are in discussions with a range of small businesses to try and ascertain their needs in this area. In the flowerpot application project, simple non-contact proximity sensors are sufficient to identify if the robot gripper is aligned correctly with the flowerpot before it is grasped. The emphasis in the project will be to develop a non-contact sensor that provides information about the object geometry.

As important as the testing of existing sensors, or the development of new sensors, is the associated real-time Digital Signal Processing (DSP) required to interpret the sensor data and translate it into meaningful information. Ultimately, the choice between different sensing solutions will probably come down to the success of the DSP in achieving the objective. A range of prototype grippers will be needed for testing purposes. The initial emphasis will be on a simple mechanical design. The use of Graspar (Crisman et al., 1996), a simple and easy to maintain mechanical structure with only 1 motor per finger provides one possibility for the initial prototype. The additional advantage of a three-finger gripper such a Graspar is that the control problem is also simplified. Additional prototypes will be developed with the intention being to move towards an industrial type intelligent gripper. Both point contact and distributed sensing solutions will be tested on the prototypes.

With regards to software LabView will be used in the development of the initial prototypes. More dedicated software developments will be required as we move towards an industrial prototype.

5. DISCUSSION

The Handyman project is a high technology project where universities are co-operating directly with industrial companies. This is a challenge, but the partners have big expectations in the outcome, especially in the area of intelligent robot handling of products. The sensory gripper technology is crucial to helping robotics become more widespread for the handling of fragile items in industry, the automation of a variety of agricultural tasks and interaction with humans in the home and at work.

6. REFERENCES

Crisman, J.B.; Kanojia,C. & Zeid, I. (1996). Graspar: A Flexible, Easily Controllable Robotic Hand, IEEE Robotics & Automation Magazine, June 1996, pp. 32-38.

Lee, M.H. & Nicholls, H.R. (1999). Tactile sensing for mechatronics--a state of the art survey, Mechatronics, Vol. 9, pp. 1-31.

Maekawa, H.; Tanie, K. & Komoriya. K. (1995). Tactile sensor based manipulation of an unknown object by a mulitfingered hand with rolling contact, Proceedings of IEEE International Conference on Robotics and Automation, pp. 743-750.

Okamura, A.M. & Cutosky, M.R. (2001). Feature-guided exploration with a robotic finger, Proceedings of the IEEE Conference on Robotics and Automation, Vol. 1, pp. 589596.

Salo, T.; Kirstein, K-U.; Vancura, T. & Baltes, H. (2007). CMOS-Based Tactile Microsensor for Medical Instrumentation, IEEE Sensors Journal, Vol. 7, No. 2, pp. 258-265, February 2007.

Shikida, M.; Shimizu, T.; Sato, K. & Itoigawa, K. (2003). Active tactile sensor for detecting contact force and hardness of an object, Sensors and Actuators A, Vol. 103, pp. 213-218.

Schmidt, P.A.; Mael, E. & R.P. Wurtz, R.P. (2006). A sensor for dynamic tactile information with applications in human-robot interaction and object exploration, Robotics and Autonomous Systems, Vol. 54, pp. 1005-1014.

Shake hands with a robot, Machine Design, pp. 88-91, February 8, 2001.

Smarter Gripping, Machine Design, January 25, 2007.

Tada, Y.; Hosada, K.; Yamasaki, Y. & Asada, M. (2003). Sensing texture of surfaces by anthropomorphic soft fingertips with multi-modal sensors, Proceedings of the IEEE/RSJ Int. Conf. on Intelligent Robots and Systems, pp. 31-35.

Tise, B. (1988). A compact high resolution piezoresistive digital tactile sensor, Proceedings of the IEEE International Conference on Robotics and Automation, Vol. 7, pp. 760-764.

Zheng, X-C.; Nakashima, R. & Yoshikawa, T. (2000). On Dynamic Control of Finger Sliding and Object Motion in Manipulation with Multifingered Hands, IEEE Transactions on Robotics and Automation, Vol. 16, No. 5, pp. 469-481, October 2000.

Zhu, F. & Spronck, J.W. (1992). A capacitive tactile sensor for shear and normal force measurements, Sensors Actuators A, Vol. 31, pp. 115-120.