Cylindrical Terrain Classification Using a Compliant Robot Foot with a Flexible Tactile-Array Sensor for Legged Robots
In this paper, we present a new approach that uses a combination of a compliant robot foot with a flexible tactile-array sensor to classify different types of cylindrical terrains. The foot and sensor were installed on a robot leg. Due to their compliance and flexibility, they can passively adapt their shape to the terrains and simultaneously provide pressure feedback during walking. We applied two different methods, which are average and maximum value methods, to classify the terrains based on the feedback information. To test the approach, We performed two experimental conditions which are (1) different diameters and different materials and (2) different materials with the same cylindrical diameter. In total, we use here eleven cylindrical terrains with different diameters and materials (i.e., a 8.2-cm diameter PVC cylinder, a 7.5-cm diameter PVC cylinder, a 5.5-cm diameter PVC cylinder, a 4.4-cm diameter PVC cylinder, a 7.5-cm diameter hard paper cylinder, a 7.4-cm diameter hard paper cylinder, a 5.5-cm diameter hard paper cylinder, a 20-cm diameter sponge cylinder, a 15-cm diameter sponge cylinder, a 7.5-cm diameter sponge cylinder, and a 5.5-cm diameter sponge cylinder). The experimental results show that we can successfully classify all terrains for the maximum value method. This approach can be applied to allow a legged robot to not only walk on cylindrical terrains but also recognize the terrain feature. It thereby extends the operational range the robot towards cylinder/pipeline inspection.
KeywordsCompliant robot foot Flexible tactile-array sensor Cylindrical terrains
This work was supported by the Capacity Building on Academic Competency of KU. Students from Kasetsart University, Thailand, Centre for BioRobotics (CBR) at University of Southern Denmark (SDU, Denmark), the Thousand Talents program of China, and the Human Frontier Science Program under grant agreement no. RGP0002/2017.
- 3.Mrva, J., Faigl, J.: Feature extraction for terrain classification with crawling robots. Robotics In: Yaghob, J. (ed.) ITAT 2015, 2010 Current and Future, pp. 179–185. Charles University in Prague, Prague (2015)Google Scholar
- 5.Degrave, J., Van Cauwenbergh, R., wyffels, F., Waegeman, T., Schrauwen, B.: Terrain classification for a quadruped robot. In: Machine Learning and Applications (ICMLA). IEEE, Miami (2013). https://doi.org/10.1109/ICMLA.2013.39
- 6.Bermudez, F.L.C., Julian, C., Haldane, W., Abbeel, P., Fearing, R. S.: Performance analysis and terrain classification for a legged robot over rough terrain. In: Intelligent Robots and Systems, IEEE/RSJ, Vilamoura (2012). https://doi.org/10.1109/IROS.2012.6386243
- 8.Kisung, K., Kwangjin K., Wansoo, K., Seungnam, Y., Changsoo H.: Performance Comparison between neural network and SVM for terrain classification of legged robot. In: SICE Annual Conference 2010, pp. 1343–1348. IEEE, Taipei (2010)Google Scholar
- 11.Walas, K.: Tactile sensing for ground classification. J. Autom. Mobile Robot. Intell. Syst. 7(2), 18–23 (2013)Google Scholar
- 13.Jamali, N., Sammut, C.: Material classification by tactile sensing using surface textures. In: Robotics and Automation, pp. 2336–2341. IEEE, Anchorage (2010). https://doi.org/10.1109/ROBOT.2010.5509675
- 14.Nakamoto, H., Kobayashi, F., Kojima, F.: Shape classification using tactile information in rotation manipulation by universal robot hand. In: Abdellatif , H. (Ed.) Robotics 2010 Current and Future Challenges. ISBN: 978-953-7619-78-7Google Scholar