SAFER vehicle inspection: a multimodal robotic sensing platform (original) (raw)

Abstract

The current threats to U.S. security both military and civilian have led to an increased interest in the development of technologies to safeguard national facilities such as military bases, federal buildings, nuclear power plants, and national laboratories. As a result, the Imaging, Robotics, and Intelligent Systems (IRIS) Laboratory at The University of Tennessee (UT) has established a research consortium, known as SAFER (Security Automation and Future Electromotive Robotics), to develop, test, and deploy sensing and imaging systems for unmanned ground vehicles (UGV). The targeted missions for these UGV systems include-but are not limited to-under vehicle threat assessment, stand-off checkpoint inspections, scout surveillance, intruder detection, obstaclebreach situations, and render-safe scenarios. This paper presents a general overview of the SAFER project. Beyond this general overview, we further focus on a specific problem where we collect 3D range scans of under vehicle carriages. These scans require appropriate segmentation and representation algorithms to facilitate the vehicle inspection process. We discuss the theory for these algorithms and present results from applying them to actual vehicle scans.

Figures (11)

Figure 1. The fundamental elements of a robotics platform. The SAFER program focuses on the processing component through fusion of multiple sensors. Figure 2. The mobile robot serves as a configurable platform that allows different sensor bricks to be added or removed. Each “SFC” brick is self contained with a sensor, fusion, and communications module.

Figure 1. The fundamental elements of a robotics platform. The SAFER program focuses on the processing component through fusion of multiple sensors. Figure 2. The mobile robot serves as a configurable platform that allows different sensor bricks to be added or removed. Each “SFC” brick is self contained with a sensor, fusion, and communications module.

Figure 3. These images depict the first prototype for the SAFER Mobile Sensor Platform. (a) The configurable sensor bay has a view portal just above the IRIS logo in this image. (b) The low-profile platform can navigate remotely under vehicles.

Figure 3. These images depict the first prototype for the SAFER Mobile Sensor Platform. (a) The configurable sensor bay has a view portal just above the IRIS logo in this image. (b) The low-profile platform can navigate remotely under vehicles.

Figure 4. Data collection with the IVP Ranger System. (a) The camera for the IVP system. (b) Color-coded range image example of a muffler. The colors indicate distances to the muffler from the camera. (c) A 3D model generated from calibration of the camera and the range image in (b).

Figure 4. Data collection with the IVP Ranger System. (a) The camera for the IVP system. (b) Color-coded range image example of a muffler. The colors indicate distances to the muffler from the camera. (c) A 3D model generated from calibration of the camera and the range image in (b).

Figure 5. Segmentation of an automotive distributor cap. (a) The original object. (b) Computer rendering of the cap after scanning into computer with IVP Ranger. (c) Segmentation of (b) into visual parts by algorithm in Page et al..’°

Figure 5. Segmentation of an automotive distributor cap. (a) The original object. (b) Computer rendering of the cap after scanning into computer with IVP Ranger. (c) Segmentation of (b) into visual parts by algorithm in Page et al..’°

Figure 6. Globally deformed superquadrics. From left to right: original object, tapered object, bent object and combi- nation of both deformations

Figure 6. Globally deformed superquadrics. From left to right: original object, tapered object, bent object and combi- nation of both deformations

Finally, considering the shape coefficients, scaling factors, coefficient for rotations, translation and the global deformations, we can describe a globally deformed superquadric positioned in space by a set of 15 parameters. where k represents the curvature of the bending plane, and a the bending angle around the z-axis. Othe intermediate parameters y, r and R are evaluated using:

Finally, considering the shape coefficients, scaling factors, coefficient for rotations, translation and the global deformations, we can describe a globally deformed superquadric positioned in space by a set of 15 parameters. where k represents the curvature of the bending plane, and a the bending angle around the z-axis. Othe intermediate parameters y, r and R are evaluated using:

Figure 7. Radial Euclidian distance between a point P and a superquadric. process heavily depend on the objective function used. Two objective functions have been primarily used in the literature. Solina and Bajcsy?° presented an objective function based on the implicit definition of superquadrics.

Figure 7. Radial Euclidian distance between a point P and a superquadric. process heavily depend on the objective function used. Two objective functions have been primarily used in the literature. Solina and Bajcsy?° presented an objective function based on the implicit definition of superquadrics.

Figure 8. Superquadric recovery process on synthetic data. (a) At initialization, ellipses are positioned into the oriented bounding box of each point cloud. At each iteration, the global fitting error is reduced. (b) After 5 iterations. (c) After 10 iterations. (d) After 20 iterations. (e) After 50 iterations. The optimization process stops when a limit number of iterations is reached or when the error is below a certain threshold.

Figure 8. Superquadric recovery process on synthetic data. (a) At initialization, ellipses are positioned into the oriented bounding box of each point cloud. At each iteration, the global fitting error is reduced. (b) After 5 iterations. (c) After 10 iterations. (d) After 20 iterations. (e) After 50 iterations. The optimization process stops when a limit number of iterations is reached or when the error is below a certain threshold.

Figure 9. This sequence of images shows the multiple scans necessary to acquire a complete image of the exhaust system The mosaic of scans is an overlay of eleven different 3D range images from the IVP Ranger.

Figure 9. This sequence of images shows the multiple scans necessary to acquire a complete image of the exhaust system The mosaic of scans is an overlay of eleven different 3D range images from the IVP Ranger.

Figure 10. View of the exhaust system underneath a vehicle. (a) Color image showing the muffler and exhaust piping. (b) Surface mesh obtained from the integration of the eleven range images. This image is a screen shot from a 3D interactive viewer.

Figure 10. View of the exhaust system underneath a vehicle. (a) Color image showing the muffler and exhaust piping. (b) Surface mesh obtained from the integration of the eleven range images. This image is a screen shot from a 3D interactive viewer.

Figure 11. Parts and superquadrics representation of the muffler. (a) Muffler and catalytic converter after cleaning, smoothing and segmentation. (b) Superquadric representation.

Figure 11. Parts and superquadrics representation of the muffler. (a) Muffler and catalytic converter after cleaning, smoothing and segmentation. (b) Superquadric representation.

Loading...

Loading Preview

Sorry, preview is currently unavailable. You can download the paper by clicking the button above.

References (26)

  1. Remotec. Andros family of tracked vehiles, Oak Ridge, TN, www.remotec-andros.com.
  2. N. S. Flann, K. L. Moore, and L. Ma, "A small mobile robot for security and inspection operations," Control Engineering Practice 10, pp. 1265-1270, Nov. 2002.
  3. iRobot. PackBot family of tracked vehiles, Burlington, MA, www.irobot.com.
  4. A. Koschan, J.-C. Ng, and M. Abidi, "Multiperspective mosaics for under vehicle inspection," in Proceedings of the SPIE Unmanned Ground Vehicle Technology, 5422, 2004.
  5. Integrated Vision Products, User Documentation: MAPP Ranger System. Sweden, 2000.
  6. A. Jaklic, A. Leonardis, and F. Solina, Segmentation and Recovery of Superquadrics, Kluwer Academic Publisher, Dordrecht, 2000.
  7. D. N. Metaxas, "Deformable models for segmentation, 3D shape and motion estimation and recognition," in Proceedings of the British Machine Vision Conference, 1999.
  8. A. P. Mangan and R. T. Whitaker, "Partitioning 3D surface meshes using watershed segmentation," IEEE Transactions on Visualization and Computer Graphics 5, pp. 308-321, Oct.-Dec. 1999.
  9. K. Wu and M. D. Levine, "3D part segmentation using simulated electrical charge distributions," IEEE Transactions on Pattern Analysis and Machine Intelligence 19, pp. 1223-1235, Nov. 1997.
  10. L. Vincent and P. Soille, "Watersheds in digital spaces: an efficient algorithm based on immersion simula- tions," IEEE Transactions on Pattern Analysis and Machine Intelligence 13, pp. 583-598, June 1991.
  11. B. Falcidieno and M. Spagnuolo, "Polyhedral surface decomposition based on curvature analysis," in Modern Geometric Computing for Visualization, T. L. Kunii and Y. Shinagawa, eds., pp. 57-72, Springer-Verlag, 1992.
  12. N. S. Sapidis and P. J. Besl, "Direct construction of polynomial surfaces from dense range images through region growing," ACM Transactions on Graphics 14, pp. 171-200, Apr. 1995.
  13. Y. Yu, A. Ferencz, and J. Malik, "Extracting objects from range and radiance images," IEEE Transactions on Visualization and Computer Graphics 7, pp. 351-364, Oct.-Dec. 2001.
  14. M. E. Rettmann, X. Han, and J. L. Prince, "Automated sulcal segmentation using watersheds on the cortical surface," NeuroImage 15, pp. 329-344, Feb. 2002.
  15. D. L. Page, A. Koschan, and M. Abidi, "Perception-based 3D triangle mesh segmentation using fast marching watersheds," in Proceedings of the International Conference on Computer Vision and Pattern Recognition, II, pp. 27-32, June 2003.
  16. D. D. Hoffman and W. A. Richards, "Parts of recognition," Cognition 18, pp. 65-96, 1984.
  17. Y. Zhang, J. K. Paik, A. Koschan, M. A. Abidi, and D. Gorsich, "A simple and efficient algorithm for part decomposition of 3d triangulated models based on curvature analysis,," in Proceedings of the International Conference on Image Processing, III, pp. 273-276, Sept. 2002.
  18. R. C. Gonzalez and R. E. Woods, Digital Image Processing, Addison-Wesley Publishing Company, Reading, MA, second ed., 2002.
  19. G. Taubin, "Estimating the tensor of curvature of a surface from a polyhedral approximation," in Proceedings of the Fifth International Conference on Computer Vision, pp. 902-907, 1995.
  20. D. L. Page, Y. Sun, A. Koschan, J. Paik, and M. Abidi, "Normal vector voting: Crease detection and curvature estimation on large, noisy meshes," Graphical Models 64, pp. 1-31, 2003.
  21. A. H. Barr, "Superquadrics and angle-preserving transformations," IEEE Transactions on Computer Graph- ics and Applications 1, pp. 11-23, Jan. 1981.
  22. A. H. Barr, "Global and local deformation of solid primitives," Computer Graphics Proceedings 18(3), pp. 21-30, 1984.
  23. M. E. Montiel, A. S. Aguado, and E. Zaluska, "Surface subdivision for generating superquadrics," Visual Computer 14(1), pp. 1-17, 1998.
  24. A. Jaklic and F. Solina, "Moments of superellipsoids and their application to range image registration," IEEE Transactions on Systems, Man, and Cybernetics, Part B 33(4), pp. 648-657, 2003.
  25. F. Solina and R. Bajcsy, "Recovery of parametric models from range images: The case for superquadrics with global deformations," IEEE Transactions on Pattern Analysis and Machine Intelligence 12(2), pp. 131-147, 1990.
  26. A. Gross and T. Boult, "Error of fit measures for recovering parametric solids," in Proceedings of the International Conference on Computer Vision, pp. 690-694, 1988. Invited paper.