Trouble designing 2D/3D LiDAR scanner
**overview of lidar project**
three of attempting build 2d lidar (light detection , ranging) system scan indoor area or tunnel system. data used reconstruct 2d image. if 2d accomplished within our timeline proceed in designing 3d system. our end goal integrate obstacle avoidance system our timeline short accomplish our focus on developing 2d/3d mapping system.
**questions of concern me**
1.)
which strategy in using laser better in generating data produce reconstructed image "map". such as....
do invert laser beam it's line reflects off objects/corners process data through edge-detection algorithm reconstruct map? https://www.youtube.com/watch?v=mz5z_xab9i8 .
do use single beam scan multiple points using "time of flight" strategy calculate distance create point cloud map. https://www.youtube.com/watch?v=mu9r_ucq2ck
2.)
3d lidar scanners extremely expensive , project requirements state 70% of project needs built. in other words cannot go out , buy scanning device.
so whatever strategy use have able make multiple scans taking in extremely large amount of data points. mean have use mcu extract , calculate individual scans , don't see happening arduino or fpga. previous sensor projects i've done i've experienced delay using xbee's transmit data computer processed.
3.)
what mcu powerful enough capture data (raspberrypi,beagleboard,arduino,fpga). need share load , transmit data computer image reconstruction. if type of transmitter/receiver without huge delay.
4.)
2d , 3d reconstruction requires heavy duty algorithms. therefore, having prebuilt library or software has built in save time.
**resources have considered using**
matlab seems used in reconstructing , mapping images through using edge-detection algorithms. there several toolboxes matlab sells deal image processing. however, don't believe use arduino,raspberrypi,beagleboard,etc matlab unless transmitted data computer processing inside matlab.
opencv located http://opencv.org/ open source c++ library "computer vision". has library full of functions image processing. compadiable on window,linux,unix systems also. however, don't think use lidar system. opencv used stereo vision. more information take http://www.intorobotics.com/fundamental-guide-for-stereo-vision-cameras-in-robotics-tutorials-and-resources/ . @ bottom of link there great tutorials explaining how use 2 webcams acquire 2d , 3d reconstructed images http://www.starlino.com/opencv_qt_stereovision.html example.
**other methods have researched project**
the overall scope of project reconstruct 2d "3d if time allows" reconstructed image "map" of entire scan start finish. though end goal incorporate obstacle avoidance system have consider method meets requirements mentioned earlier. other methods i've considered follows.....
sonar https://www.youtube.com/watch?v=u-mxsgfw6qs&list=ple32ffada68d35f72
stereo vision https://www.youtube.com/watch?v=xghb85ccky8
3d laser mapping processing data "point cloud" https://www.youtube.com/watch?v=bicn3hucut4
i'm not sure video demonstrating in whether using lidar or sonar. it's using slam "simultaneous localization , mapping" alogrithms generate 3d map. seems theclosest idea trying simulate https://www.youtube.com/watch?v=quqf5_ze_fi.
**advice & guidance**
as seen sonar method looks nice , match end goal of our project. however, stereo vision allow me use opencv free , open source on platform raspberrypi or beagleboard. technically use matlab stereo vision. however, lidar seems promising , require more hardware design meet requirements more.
any guidance/information has me on find more information or advice on direction should take extremely helpful , appreciated. our current funding our project @ $3000. going approval board here in 2 weeks , in group having hard time deciding on direction more approachable. therefore, hearing other input useful @ point.
three of attempting build 2d lidar (light detection , ranging) system scan indoor area or tunnel system. data used reconstruct 2d image. if 2d accomplished within our timeline proceed in designing 3d system. our end goal integrate obstacle avoidance system our timeline short accomplish our focus on developing 2d/3d mapping system.
**questions of concern me**
1.)
which strategy in using laser better in generating data produce reconstructed image "map". such as....
do invert laser beam it's line reflects off objects/corners process data through edge-detection algorithm reconstruct map? https://www.youtube.com/watch?v=mz5z_xab9i8 .
do use single beam scan multiple points using "time of flight" strategy calculate distance create point cloud map. https://www.youtube.com/watch?v=mu9r_ucq2ck
2.)
3d lidar scanners extremely expensive , project requirements state 70% of project needs built. in other words cannot go out , buy scanning device.
so whatever strategy use have able make multiple scans taking in extremely large amount of data points. mean have use mcu extract , calculate individual scans , don't see happening arduino or fpga. previous sensor projects i've done i've experienced delay using xbee's transmit data computer processed.
3.)
what mcu powerful enough capture data (raspberrypi,beagleboard,arduino,fpga). need share load , transmit data computer image reconstruction. if type of transmitter/receiver without huge delay.
4.)
2d , 3d reconstruction requires heavy duty algorithms. therefore, having prebuilt library or software has built in save time.
**resources have considered using**
matlab seems used in reconstructing , mapping images through using edge-detection algorithms. there several toolboxes matlab sells deal image processing. however, don't believe use arduino,raspberrypi,beagleboard,etc matlab unless transmitted data computer processing inside matlab.
opencv located http://opencv.org/ open source c++ library "computer vision". has library full of functions image processing. compadiable on window,linux,unix systems also. however, don't think use lidar system. opencv used stereo vision. more information take http://www.intorobotics.com/fundamental-guide-for-stereo-vision-cameras-in-robotics-tutorials-and-resources/ . @ bottom of link there great tutorials explaining how use 2 webcams acquire 2d , 3d reconstructed images http://www.starlino.com/opencv_qt_stereovision.html example.
**other methods have researched project**
the overall scope of project reconstruct 2d "3d if time allows" reconstructed image "map" of entire scan start finish. though end goal incorporate obstacle avoidance system have consider method meets requirements mentioned earlier. other methods i've considered follows.....
sonar https://www.youtube.com/watch?v=u-mxsgfw6qs&list=ple32ffada68d35f72
stereo vision https://www.youtube.com/watch?v=xghb85ccky8
3d laser mapping processing data "point cloud" https://www.youtube.com/watch?v=bicn3hucut4
i'm not sure video demonstrating in whether using lidar or sonar. it's using slam "simultaneous localization , mapping" alogrithms generate 3d map. seems theclosest idea trying simulate https://www.youtube.com/watch?v=quqf5_ze_fi.
**advice & guidance**
as seen sonar method looks nice , match end goal of our project. however, stereo vision allow me use opencv free , open source on platform raspberrypi or beagleboard. technically use matlab stereo vision. however, lidar seems promising , require more hardware design meet requirements more.
any guidance/information has me on find more information or advice on direction should take extremely helpful , appreciated. our current funding our project @ $3000. going approval board here in 2 weeks , in group having hard time deciding on direction more approachable. therefore, hearing other input useful @ point.
any guidance/information has me on find more information or advice on direction should take extremely helpful , appreciated. our current funding our project @ $3000. going approval board here in 2 weeks , in group having hard time deciding on direction more approachable. therefore, hearing other input useful @ point.
there opencv mailing list can ask feedback feasibility of image processing you're proposing.
if/when know basic approach feasible can start thinking resources it's going need. i'm sure you'll find microcontroller not suitable platform this. unless have hidden requirement portable, suggest implement on decent spec computer. if have obscure reason wanting portable solution, there mini computers such rpi consider. standard pc easier , better though.
Arduino Forum > Using Arduino > Project Guidance > Trouble designing 2D/3D LiDAR scanner
arduino
Comments
Post a Comment