An Image Based Method of Finding Better Walking Strategies for Hexapod on Discontinuous Terrains
Download as PPT, PDF0 likes438 views
The document proposes an image-based method for finding better walking strategies for a hexapod robot on discontinuous terrains. It simulates a terrain environment, develops a hexapod walking model with different gaits, and defines an intelligent walking strategy. The strategy selects the optimal gait based on terrain analysis from images and parameters like stride length. It enables the hexapod to traverse rugged terrain by adapting its gait selection and switching between tripod and symmetrical walking patterns when needed.
![ Measured ranges of movement (Measured ranges of movement (KKmm) for various angle &) for various angle &
stored in the matrix.stored in the matrix.
The largest value defined asThe largest value defined as 255255 (white), lowest as(white), lowest as 00 (black).(black).
This study chooseThis study choose 8 forward gait angle8 forward gait angle matrix databasematrix database
[0 °, 15 °, 30 °, 45 °, 60 °, -15 °, -30 °, -45 °][0 °, 15 °, 30 °, 45 °, 60 °, -15 °, -30 °, -45 °]
12](https://image.slidesharecdn.com/aimkaziv2-160211065345/85/An-Image-Based-Method-of-Finding-Better-Walking-Strategies-for-Hexapod-on-Discontinuous-Terrains-12-320.jpg)
An Image Based Method of Finding Better Walking Strategies for Hexapod on Discontinuous Terrains
- 1. An Image Based Method of Finding Better WalkingAn Image Based Method of Finding Better Walking Strategies for Hexapod on DiscontinuousTerrainsStrategies for Hexapod on DiscontinuousTerrains Kazi MostafaKazi Mostafa kazimostafa@gmail.com Professor Innchyn HerProfessor Innchyn Her her@mail.nsysu.edu.tw Kung-Ting WeiKung-Ting Wei weikt321@gmail.com Department of Mechanical & Electro-Mechanical EngineeringDepartment of Mechanical & Electro-Mechanical Engineering National Sun Yat-sen University, TaiwanNational Sun Yat-sen University, Taiwan
- 2. IntroductionIntroduction ◦ Previous Research WorkPrevious Research Work ◦ ObjectivesObjectives MethodsMethods ◦ Simulate Terrain EnvironmentSimulate Terrain Environment ◦ Proposed Hexapod ModelProposed Hexapod Model ◦ Walking StrategyWalking Strategy ResultsResults ConclusionConclusion 2
- 3. Wheeled vs.Wheeled vs. LeggedLegged locomotion.locomotion. OnOn discontinuous terrainsdiscontinuous terrains.. Multi legged animals have excellentMulti legged animals have excellent adaptabilityadaptability to traverse overto traverse over rugged terrainrugged terrain.. Multi-leggedMulti-legged robots that imitate crawling insects.robots that imitate crawling insects. HexapodHexapod configuration is mostconfiguration is most efficient &efficient & statically stablestatically stable.. 3
- 4. 4
- 5. Vision systemVision system to recognize surroundingto recognize surrounding terrainterrain Image based methodImage based method of finding walkingof finding walking strategiesstrategies To simulateTo simulate a completea complete hexapod walkinghexapod walking modelmodel 5
- 6. Proposed research divided into 3 major steps:Proposed research divided into 3 major steps: A.A.Establishment of Terrain EnvironmentEstablishment of Terrain Environment B.B.The Hexapod Gait ModelThe Hexapod Gait Model - A simplified forward gait- A simplified forward gait - Rotation- Rotation - Intelligent gait control- Intelligent gait control A.A.Hexapod’s Walking StrategyHexapod’s Walking Strategy -- Gait selection methodGait selection method - Walking Strategy- Walking Strategy 6
- 7. Proposed research divided into 3 major steps:Proposed research divided into 3 major steps: A.A.Establishment of Terrain EnvironmentEstablishment of Terrain Environment B.B.The Hexapod Gait ModelThe Hexapod Gait Model - A simplified forward gait- A simplified forward gait - Rotation- Rotation - Intelligent gait control- Intelligent gait control A.A.Hexapod’s Walking StrategyHexapod’s Walking Strategy -- Gait selection methodGait selection method - Walking Strategy- Walking Strategy 7
- 8. DefinedDefined sizesize of the terrain & entered 250 circles.of the terrain & entered 250 circles. ComputerComputer displayeddisplayed a topographic vector map bya topographic vector map by using those circles.using those circles. AppliedApplied Grayscale erosionGrayscale erosion to find the boundaryto find the boundary of Circles.of Circles. 8
- 9. The white regions areThe white regions are standingstanding areas, black regions areareas, black regions are forbiddenforbidden areas & the yellow region is being eroded.areas & the yellow region is being eroded. TheThe edgesedges of the terrain changes into a forbidden area.of the terrain changes into a forbidden area. 9 Foothold Zone Forbidden Zone Forbidden Zone
- 10. Proposed research divided into 3 major steps:Proposed research divided into 3 major steps: A.A.Establishment of Terrain EnvironmentEstablishment of Terrain Environment B.B.The Hexapod Gait ModelThe Hexapod Gait Model - A simplified forward gait- A simplified forward gait - Rotation- Rotation - Intelligent gait control- Intelligent gait control A.A.Hexapod’s Walking StrategyHexapod’s Walking Strategy -- Gait selection methodGait selection method - Walking Strategy- Walking Strategy 10
- 11. Semi-circular shaped areas are Hexapod gait range of motionSemi-circular shaped areas are Hexapod gait range of motion Maximum range RMaximum range Rmaxmax = 60 (pixels), minimum range R= 60 (pixels), minimum range Rminmin = 20= 20 (pixels), length between two successive gaits B = 100(pixels), length between two successive gaits B = 100 (pixels), width W = 50 (pixels), and the offset value D = 50(pixels), width W = 50 (pixels), and the offset value D = 50 (pixels).(pixels). 11
- 12. Measured ranges of movement (Measured ranges of movement (KKmm) for various angle &) for various angle & stored in the matrix.stored in the matrix. The largest value defined asThe largest value defined as 255255 (white), lowest as(white), lowest as 00 (black).(black). This study chooseThis study choose 8 forward gait angle8 forward gait angle matrix databasematrix database [0 °, 15 °, 30 °, 45 °, 60 °, -15 °, -30 °, -45 °][0 °, 15 °, 30 °, 45 °, 60 °, -15 °, -30 °, -45 °] 12
- 13. Parameters of forward gait:Parameters of forward gait: forward angle , strideαforward angle , strideα length S, and Slength S, and SXX, S, SYY are the S in X direction and Y direction.are the S in X direction and Y direction. Various parameters of forward gait 13 α -15° -30° -45° 0° 15° 30° 45° 60° S 111.6 101.2 83.4 112 111.6 101.2 83.4 67.2 SX -28 -50 -59 0 28 50 59 58 SY 108 88 59 112 108 88 59 34
- 14. Measurement of gait rotation can be divided into twoMeasurement of gait rotation can be divided into two steps;steps; FirstFirst, we calculated the distance between each point, we calculated the distance between each point of motion range and C.G.of motion range and C.G. SecondSecond, used same size of matrix to store the range of, used same size of matrix to store the range of movement .movement . 14 Rotation Angle θ2 θ3 θ6
- 15. When the hexapod robot rotates clockwise, theWhen the hexapod robot rotates clockwise, the center ofcenter of rotationrotation move to right direction.move to right direction. The value of θThe value of θ33 decreases & θdecreases & θ22,θ,θ66 increases gradually.increases gradually. When θWhen θ22, θ, θ33, θ, θ66 turnedturned equalequal, hexapod could get, hexapod could get maximummaximum rotation anglerotation angle.. The location of OThe location of OTT isis 6262 units right from CGunits right from CG 15 62 Units
- 16. The rotation around CG hasThe rotation around CG has nono forward movement.forward movement. But, for the maximum rotation angle of 37.7° gait, theBut, for the maximum rotation angle of 37.7° gait, the robot can move forward ofrobot can move forward of 40.2 units40.2 units, which includes, which includes 13 units in X direction, 38 units Y direction.13 units in X direction, 38 units Y direction. 16 θ S SX SY R Rotation around C.G. 30.03° 0 0 0 0 Max. angle of rotation 37.67° 40.1640.16 1313 3838 6262
- 17. This method assumes that the target point is in frontThis method assumes that the target point is in front of the robot & center of rotation is in the right rear.of the robot & center of rotation is in the right rear. In each step robot can rotate degree depends onθIn each step robot can rotate degree depends onθ the condition and distance of movement.the condition and distance of movement. After finished movement, hexapod’s center ofAfter finished movement, hexapod’s center of rotation can be measured.rotation can be measured. Then, by using previously mentioned method,Then, by using previously mentioned method, measured the radius of rotation and parameters formeasured the radius of rotation and parameters for 5° and 10° of rotation.5° and 10° of rotation. 17
- 18. Stride length S for 5° is 101.9 unit and for 10° is 93 unit.Stride length S for 5° is 101.9 unit and for 10° is 93 unit. For10°, the hexapod moved less distance in the Y direction,For10°, the hexapod moved less distance in the Y direction, but it added value in rotation.but it added value in rotation. If Hexapod needs to move & want to rotate simultaneously,If Hexapod needs to move & want to rotate simultaneously, it should choose 5° of rotational gait.it should choose 5° of rotational gait. If the greater rotation is needed, hexapod might be choosingIf the greater rotation is needed, hexapod might be choosing 10°.10°. 18 θ S SX SY R 5° 101.92 28 98 1162.3 10° 93.01 27 89 527.68
- 19. First, take snapshot & pre-calculated the next foothold point.First, take snapshot & pre-calculated the next foothold point. Based on the terrain's ranges overlap with the gait ranges ofBased on the terrain's ranges overlap with the gait ranges of motion, themotion, the 33rdrd gaitgait is beyond the region.is beyond the region. So the robot chooses 1-4-5 gait sequence for foothold. AndSo the robot chooses 1-4-5 gait sequence for foothold. And select the maximum number of points (gray region).select the maximum number of points (gray region). 19 Terrain’s Foothold Zone Gait Ranges of Motion OverlappedNo Foothold 1 4 5
- 20. Started with tripod gait. Difficulties in transformation between 1-4-5 andStarted with tripod gait. Difficulties in transformation between 1-4-5 and 2-3-6 gait. Gait number 2 had no foothold.2-3-6 gait. Gait number 2 had no foothold. Hexapod using symmetrical gait (1-3-4-6) for foothold and move.Hexapod using symmetrical gait (1-3-4-6) for foothold and move. Hexapod again converted back in to tripod gait (2-3-6). Thus, if hexapodHexapod again converted back in to tripod gait (2-3-6). Thus, if hexapod stuck again it will transformed 1-4-5 gait sequence to continue walk.stuck again it will transformed 1-4-5 gait sequence to continue walk. 20 Current Km Residual Km 1 3 4 6 No Foothold
- 21. 211-4-5 Gait Sequence 2-3-6 Gait Sequence Symmetrical Gait Target
- 22. Proposed research divided into 3 major steps:Proposed research divided into 3 major steps: A.A.Establishment of Terrain EnvironmentEstablishment of Terrain Environment B.B.The Hexapod Gait ModelThe Hexapod Gait Model - A simplified forward gait- A simplified forward gait - Rotation- Rotation - Intelligent gait control- Intelligent gait control A.A.Hexapod’s Walking StrategyHexapod’s Walking Strategy -- Gait selection methodGait selection method - Walking Strategy- Walking Strategy 22
- 23. Proposed hexapod has 8 different angles of forward gait. InProposed hexapod has 8 different angles of forward gait. In total 64 combinations of gaits. But, 0° forward gait is a firsttotal 64 combinations of gaits. But, 0° forward gait is a first choice.choice. Categorized these 64 combination of gait into 3 groups.Categorized these 64 combination of gait into 3 groups. The combination of smaller angle gait is the most preferredThe combination of smaller angle gait is the most preferred one.one. In every step, hexapod compared stride length with TableIn every step, hexapod compared stride length with Table 11. If. If stride length is at least 80% of tablestride length is at least 80% of table 11, hexapod walk by 0°, hexapod walk by 0° forward gait.forward gait. If it is less, hexapod will choose other optimum gait angle.If it is less, hexapod will choose other optimum gait angle. 23 α -15° -30° -45° 0° 15° 30° 45° 60° S 111.6 101.2 83.4 112 111.6 101.2 83.4 67.2 SX -28 -50 -59 0 28 50 59 58 SY 108 88 59 112 108 88 59 34 Table 1
- 24. Group 01Group 01 0°0° 0°, 15°, -15°0°, 15°, -15° 15°15° 30°, 0°30°, 0° -15°-15° 30°, 0°30°, 0° Measure StrideMeasure Stride LengthLength StrideStride length islength is 80% of max80% of max stridestride KeepKeep MovingMoving Group 02Group 02 0°0° 30°, -30°30°, -30° 15°15° 45°, -15°45°, -15° -15°-15° -45°, 15°-45°, 15° 30°30° 0°, 15°, 45°0°, 15°, 45° -30°-30° 0°,-15°,-45°0°,-15°,-45° Measure StrideMeasure Stride LengthLength StrideStride length islength is 80% of max80% of max stridestride KeepKeep MovingMoving YesYes NoNo … NoNoYesYes
- 25. The walking strategy database contains 8 kinds of linearThe walking strategy database contains 8 kinds of linear motion, 4 rotational movements, and the tripod gait andmotion, 4 rotational movements, and the tripod gait and the symmetrical gait.the symmetrical gait. The hexapod’s first choice is 0 ° forward gait for largestThe hexapod’s first choice is 0 ° forward gait for largest stride length and ± 10 ° is second.stride length and ± 10 ° is second. If the angle between the robot and the target is greaterIf the angle between the robot and the target is greater than 45°, then hexapod choose the rotation gait.than 45°, then hexapod choose the rotation gait. If the angle between the robot and the target is less thanIf the angle between the robot and the target is less than 45 °, then 5 ° or 10 ° of rotation gait can be used.45 °, then 5 ° or 10 ° of rotation gait can be used. 25
- 26. 26 Angle (α) of Target Less than 5° 0° Straight Forward Gait Next step foothold Yes Hexapod moving to target Reaches Target Point Yes Complete No Greater than 5° Angle (α) of Target Greater than 45° Max angle of rotation Less than 45° Rotation around any point Angle (α) of Target Greater than 5°Less than 5° Transform to symmetrical gait No Next step foothold Yes
- 27. IntroductionIntroduction ◦ Previous Research WorkPrevious Research Work ◦ ObjectivesObjectives MethodsMethods ◦ Simulate Terrain EnvironmentSimulate Terrain Environment ◦ Hexapod ModelHexapod Model ◦ Walking StrategyWalking Strategy ResultsResults ConclusionConclusion 27
- 28. In proposed simulated terrain, there are 250 circlesIn proposed simulated terrain, there are 250 circles with radius 15, 30, 45, 60, 75 where hexapod canwith radius 15, 30, 45, 60, 75 where hexapod can stand. The average radius of these circles is 29.7.stand. The average radius of these circles is 29.7. The significant advantage of proposed hexapod isThe significant advantage of proposed hexapod is use of databases with image processing technique.use of databases with image processing technique. Hexapod can use database to reduce computationHexapod can use database to reduce computation time.time. Proposed design is based on KProposed design is based on Kmm to select gait andto select gait and foothold.foothold. 28
- 29. 29
- 30. IntroductionIntroduction ◦ Previous Research WorkPrevious Research Work ◦ ObjectivesObjectives MethodsMethods ◦ Simulate Terrain EnvironmentSimulate Terrain Environment ◦ Hexapod ModelHexapod Model ◦ Walking StrategyWalking Strategy ResultsResults ConclusionConclusion 30
- 31. This study showed the use of image processingThis study showed the use of image processing based technique to design an adaptive gait.based technique to design an adaptive gait. Matrix database combined with image processingMatrix database combined with image processing method made the hexapod’s walking strategy easymethod made the hexapod’s walking strategy easy to determine.to determine. Thus, the robot can simulate routes in advance, andThus, the robot can simulate routes in advance, and achieve the artificial intelligence capabilities.achieve the artificial intelligence capabilities. Finally, we hope that the proposed research methodFinally, we hope that the proposed research method and strategy can provide a reference for futureand strategy can provide a reference for future research to scholars.research to scholars. 31
- 32. As part of our continuing research, we areAs part of our continuing research, we are developing a physical model by using proposeddeveloping a physical model by using proposed computer simulation results and improvingcomputer simulation results and improving efficiency for real-time control application toefficiency for real-time control application to proof the optimality of the design.proof the optimality of the design. 32
Editor's Notes
- #21: The figure shows that the green circle is 1-4-5 gait sequence, and the blue box is 2-3-6 gait sequence.
- #22: Green circles are 1-4-5 gait sequence, and the blue boxes are 2-3-6 gait sequence. Purple triangle is 3 & 6 gait
- #27: After Create terrain and set target point
- #30: Robots walking record in rough terrain is shown in Fig.16.