a robot trajectory programming method using multi camera
play

A robot trajectory programming method using multi-camera systems - PowerPoint PPT Presentation

A robot trajectory programming method using multi-camera systems SILVIO GIANCOLA DAVIDE CHIARION REMO SALA MESA2014 SENIGALLIA 11/09/2014 2 Introduction Study of a need : Tiles decoration SUMMARY Hand made Serigraphy Introduction


  1. A robot trajectory programming method using multi-camera systems SILVIO GIANCOLA DAVIDE CHIARION REMO SALA MESA2014 – SENIGALLIA – 11/09/2014

  2. 2 Introduction Study of a need : Tiles decoration SUMMARY Hand made Serigraphy Introduction • • Tool Identification Authenticity and unicity Fast method Advantages • Trajectory Smoothing High quality product High quantity • Metrologic Analysis Poor quantity Poor value Application • Inconvenient Necessity of a qualified person Printed Conclusion • Innovative decoration process? Ø Use of anthropomorphic robot for the tile painting Robot programming method? Ø Paintbrush trajectory registration A robot trajectory programming method using multi-camera systems

  3. 3 Tool Identification SUMMARY Introduction • • Tool Identification – Stereoscopic System Point Cloud – Indexing – Rigid Body Registration AMBIGUITY • Trajectory Smoothing LIGHT NON CONDITION Metrologic Analysis • INTRUSIVE Application • • Conclusion OCCLUSION PRECISION A robot trajectory programming method using multi-camera systems

  4. 4 Tool Identification SUMMARY Introduction • • Tool Identification – Stereoscopic System Point Cloud – Indexing – Rigid Body Registration AMBIGUITY • Trajectory Smoothing LIGHT CONDITION Metrologic Analysis • Application • • Conclusion OCCLUSION A robot trajectory programming method using multi-camera systems

  5. 5 Tool Identification SUMMARY Introduction • • Tool Identification – Stereoscopic System Point Cloud – Indexing – Rigid Body Registration • Trajectory Smoothing Metrologic Analysis • Application • • Conclusion A robot trajectory programming method using multi-camera systems

  6. 6 Tool Identification Stereoscopic system § Based on 2 cameras SUMMARY Introduction • • Tool Identification – Stereoscopic System Point Cloud – Indexing – Rigid Body Registration • Trajectory Smoothing Metrologic Analysis • Application • • Conclusion A robot trajectory programming method using multi-camera systems

  7. 7 Tool Identification Trinocular stereoscopy SUMMARY Introduction • • Tool Identification – Stereoscopic System Point Cloud – Indexing – Rigid Body Registration • Trajectory Smoothing Metrologic Analysis • Application • • Conclusion A robot trajectory programming method using multi-camera systems

  8. 8 Tool Identification Trajectory Acquisition from Trinocular Vision System § Point Cloud Indexing SUMMARY Introduction • § Rigid Body Registration • Tool Identification – Stereoscopic § Trajectory Smoothing System Point Cloud – Indexing – Rigid Body Registration 1 4 • Trajectory Smoothing 2 3 Metrologic Analysis • Application • • Conclusion 3 1 4 2 A robot trajectory programming method using multi-camera systems

  9. 9 Tool Identification Point Cloud Ordering § Experimental method using distances matrix SUMMARY Introduction • • Tool Identification – Stereoscopic System Point Cloud – Indexing – Rigid Body § Ex: Registration • Trajectory Smoothing Metrologic Analysis • Application • • Conclusion A robot trajectory programming method using multi-camera systems

  10. 10 Tool Identification Point Cloud Ordering – Algorithm § If n is the number of marker, we split the n x n matrix in n rows. Be row i the i th row. SUMMARY Introduction • § For each row i , we define an empty 2 x n array named ind i , linked to the i th row. • Tool Identification – Stereoscopic § For each element d ij of row i , its value is matched to a correspondent value d kl of the System model. The rows and columns indexes (k and l) are then saved in the columns of Point Cloud – ind i . Indexing – Rigid Body Registration § For each ind i , we define i sol the largest occurrence of an index in this matrix. • Trajectory Smoothing Metrologic Analysis § If all the element of ind i is 0, then a marker is missing. • Application • § Else, the largest occurrence of an index different to 0 returns the correct index • Conclusion of i. A robot trajectory programming method using multi-camera systems

  11. 11 Tool Identification 0 𝑒13 𝑒15 𝑒12 𝑌 Point Cloud Ordering – Example 𝑒31 0 𝑒35 𝑒32 𝑌 𝑒51 𝑒53 0 𝑒52 𝑌 𝑒21 𝑒23 𝑒25 0 𝑌 SUMMARY Introduction • 𝑌 𝑌 𝑌 𝑌 0 • Tool Identification § ind 1 = 0 1 1 1 0 0 , the 1st marker corresponds to the 1st one – Stereoscopic 0 3 5 2 System § ind 2 = 3 0 3 3 0 0 , the 2nd marker corresponds to the 3rd one Point Cloud – 1 0 5 2 Indexing – Rigid Body § ind 3 = 5 5 0 5 0 0 , the 3rd marker corresponds to the 5th one Registration 1 3 0 2 • Trajectory Smoothing § ind 4 = 2 2 2 0 0 0 , the 4th marker corresponds to the 2nd one Metrologic Analysis • 1 3 5 0 Application • 0 , a marker is missing (the 4th). § ind 5 = 0 0 0 0 0 • Conclusion 0 0 0 0 A robot trajectory programming method using multi-camera systems

  12. 12 Tool Identification Rigid Body Registration § Find Rotation and Translation of a point cloud respect to a model SUMMARY Introduction • • Tool Identification – Stereoscopic System Point Cloud – Indexing – Rigid Body Registration • Trajectory Smoothing Metrologic Analysis • Application • • Conclusion A robot trajectory programming method using multi-camera systems

  13. 13 Tool Identification Rigid Body Registration § Find Rotation Translation and Scaling of a point cloud respect to a model SUMMARY Introduction • § Singular Value Decomposition (SVD) • Tool Identification - ∗ 𝑠 ,/ = 𝑉 ∗ 𝑇 ∗ 𝑊 – Stereoscopic 3 § 𝑁 = ∑ 𝑠 , ,45 , System - and 𝑠 w 𝑗𝑢ℎ 𝑠 , the deviation of points of a cloud respect to its barycenter , Point Cloud – 1 0 0 Indexing ∗ 𝑊 / 0 1 0 § 𝑆𝑝𝑢𝑏𝑢𝑗𝑝𝑜 = 𝑉 ∗ – Rigid Body 𝑡𝑗𝑕𝑜 𝑒𝑓𝑢 𝑉 ∗ 𝑊 / 0 0 Registration • Trajectory Smoothing § 𝑈𝑠𝑏𝑜𝑡𝑚𝑏𝑢𝑗𝑝𝑜 = 𝑞̂ - − 𝑆𝑝𝑢𝑏𝑢𝑗𝑝𝑜 ∗ 𝑞̂ Metrologic Analysis • w 𝑗𝑢ℎ 𝑞̂ - 𝑓𝑢 𝑞̂ the barycenter of the point clouds Application • • Conclusion A robot trajectory programming method using multi-camera systems

  14. 14 Trajectory Smoothing Filtering and Interpolations : § Cubic / Spline interpolation SUMMARY § Bézier Curves interpolations Introduction • § B-Spline interpolations (NURBS) • Tool Identification • Trajectory Smoothing § Low Pass filter • Metrologic Analysis Application • Conclusion Tests : • § Sinus Following § White Noise reduction § Sinus and White Noise Combination § Acquired Data A robot trajectory programming method using multi-camera systems

  15. 15 Trajectory Smoothing Results : SUMMARY Sinusoids White noise Combination Acquired data Introduction • • Tool Identification ++ -- -- - Cubic / Spline • Trajectory Smoothing • Metrologic Analysis -- ++ - - Application • Bézier curves Conclusion • + + + + B-Spline (NURBS) + + + + Low Pass A robot trajectory programming method using multi-camera systems

  16. 16 Trajectory Smoothing Amplitude Spectrum of an acquisition: SUMMARY Introduction • • Tool Identification • Trajectory Smoothing • Metrologic Analysis Application • Conclusion • A robot trajectory programming method using multi-camera systems

  17. 17 Metrologic Analysis Metrologic Analysis SUMMARY Introduction • • Accuracy (AP) : proximity of • Tool Identification measurement results to the true • Trajectory Smoothing value • Metrologic Analysis Application • Conclusion • • Repeatibility (RP) : precision of the measurement A robot trajectory programming method using multi-camera systems

  18. 18 Metrologic Analysis Metrologic Analysis § Repeatability SUMMARY § Analysis of the reconstruction of a marker at the origin position Introduction • § Statistical study (gaussian) • Tool Identification • Trajectory Smoothing § Results • Metrologic Analysis § Uncertainty respect to the axis : Application • Conclusion • § Y axis : 0,05 mm § X axis : 0,21 mm § Z axis : 0,15 mm A robot trajectory programming method using multi-camera systems

  19. 19 Metrologic Analysis Metrologic Analysis § Accuracy SUMMARY § Trajectory analysis among the coordinate system axis Introduction • § Use of the Rigid Registration algorithm on the point cloud defining the • Tool Identification trajectory • Trajectory Smoothing • Metrologic Analysis Application • § Results Conclusion • § Error among X camera axis: 0.7 mm § Error among Y camera axis: 1.2 mm § Error among Z camera axis: 3.7 mm A robot trajectory programming method using multi-camera systems

  20. 20 Application Application § Model of the robot : ABB IRB 120 SUMMARY Introduction • • Tool Identification • Trajectory Smoothing • Metrologic Analysis Application • Conclusion • A robot trajectory programming method using multi-camera systems

  21. 21 Application Application SUMMARY Introduction • • Tool Identification • Trajectory Smoothing • Metrologic Analysis Application • Conclusion • A robot trajectory programming method using multi-camera systems

Recommend


More recommend