NVIDIA JetBot§Ö³t´x´¤AIÃä½t­pºâ­n»â

¥»½Òµ{¾Ç­û¥ç¥i¥[¤JVIP¤¬§U¦¡¾Ç²ß°Q½×¸s¡A§K¶O¨É¨üJetBotÁ¿®v»P¦P.....

¾A¦X¹ï¶H¡G

NVIDIA JetBot§Ö³t´x´¤AIÃä½t­pºâ­n»â [ªê§À³õ]

 

¤W½Ò¦a§}¡G°ê¥ßªê§À¬ì§Þ¤j¾Ç ²Ä¤G±Ð¾Ç°Ï²Ä¤Tºî¦X¤uµ{À]2¼Ó STEAM¹ê§@±Ð«Ç (¶³ªL¿¤ªê§ÀÂí¤å¤Æ¸ô64¸¹) ¹ê»Ú¤W½Ò¦aÂI¡A½Ð¨Ì¤W½Ò³qª¾¬°·Ç!

®É¼Æ¡G7

°_¨´¤é´Á¡G2019-12-27~2019-12-27

Ápµ¸¸ê°T¡GªLÁâ¯ø/07-3367833 Âà14

³ø¦WºI¤î¤é¡G2019-12-26

½Òµ{Ãþ§O¡G¤H¤~°ö°V(½Òµ{)

¬¡°Ê¥N½X¡G2319100044


½Òµ{¤¶²Ð

§ó§Ö³t¡B§ó²³æªº¶}µo»P³¡¸pAIÃä½t­pºâ²£«~ªº¤èªk

§@爲NVIDIA®­«ÂI±À¼sªºJetbotÃä½t­pºâ¾Ç²ß¥­¥x¡A¥H¥þ²y­º´Ú¦P®É¸Ñ¨M©Ê¯à¡BÅé¿n¡B¥\¯Ó»P»ù®æµ¥¥|¤j°ÝÃDªºJetson Nanoºî¦X©ÊAIªä¤ù¬°®Ö¤ß¡A¹w¥ý¦w¸ËÂ×´IªºDeep Learning°t®M¶}µo¸ê·½¡]¥]¬ATensorFlow¡BPytorch¡BKeras¡BYoloµ¥¡^¥H¤Î»P¾ð²ù¬£§¹¥þ¬Û®eªº40-pin±µ¸}ªºAPI¡]AdafruitªºGPIO®w¡^¡A¹ý©³±N“´¼¡]²`«×¾Ç²ß¡^”»P“¯à¡]¾÷¹q±±¨î¡^”§¹¬üµ²¦X©ó¤@¨­¡A爲¶}µoªÌÁYµu80%¥H¤W©³¼h»P©PÃä³]³Æªº¶}µo®É¶¡»P¦¨¥»¡A¦¨爲³Ì§Ö³t¶}µoÃä½t­pºâ³]³Æªº¾Ç²ß¥­¥x¡C

JetBot´¼¯à¤p¨®¥HÀu¶®³y«¬¡B²¤ÆªºµwÅé³]³Æµ²ºc¡B¥­©öªñ¤HªºJupyter ¤¬°Ê¬É­±¡A·f«Ø³Ì»´©ö¤W¤âªºÃä½t­pºâ¾Ç²ß¥­¥x¡C¥»½Òµ{¸Ô²Ó»¡©ú»P¾Þ§@¦p¦ó±N²`«×¾Ç²ß¡]Deep Learning¡^¡B¾E²¾¾Ç²ß¡]Transfer Learning¡^§Þ³NÀ³¥Î¨ìJetbot¦Û°Ê¾Þ±±¤W¡A§¹¦¨¦h­Ó¥i¥H¥ß§Y¹ê½îªº§¹¾ãÀ³¥Î¡A¥]¬A¥Ø¼Ðª«¸òÀH¡]Object Following¡^¡B¸úÁ×»Ùê¡]Collision Avoidance¡^¡BÀH¾÷´M§ä¸ô®|¡]Find Path¡^µ¥”¼ÒÀÀµL¤H¾r¾p“ªº¹ê§@½m²ß¡A¾Ç­û¯à»´ÃP´x´¤§¹¾ãªºAIÃä½t­pºâª¾ÃѹÏÃСA§Ö³t¤JªùAI»â°ì¡BÅéÅç²`«×¾Ç²ß¹ê¾Ô¬yµ{¡B¤F¸Ñ´¼¯à¤p¨®³]­p°ò¦ª¾ÃÑ¡A¤Îºc«Ø¦Û¤vªº²`«×¾Ç²ß¡]¥]¬A¦Û°Ê¾r¾p¡B¾÷±ñ¤âÁu±±¨î¡^¼Ò«¬¡C

»²§UJetBot½Òµ{³¡¤ÀÁÙ´£¨Ñ”²{³õÀò±oNVIDIA Jetson Nano¾Ç²ßÃҮѓªº«ü¾É¡A¥H¤ÎTensorRT 5.1.6¥[³t¤ÞÀº¡BDeepStream 4.0µøÀW¤ÀªR³nÅ骺¨Ï¥Î»PÀu¤ÆÁ¿¸Ñ¡A¾Ç²ß¤§«á¥i¥H¦Û¦æ®Ú¾Ú©Ò»Ý¶i¦æ²`«×½Õ¾ã¡C¥»½Òµ{¾Ç­û¥ç¥i¥[¤JVIP¤¬§U¦¡¾Ç²ß°Q½×¸s¡A§K¶O¨É¨üJetBotÁ¿®v»P¦P¦n±M®a­Ì¬ÛÃöªº§Þ³N¿Ô¸ß¡C



 
 
½Òµ{¯S¦â/¥Ø¼Ð
  ½Òµ{¥Ø¼Ð
³q¹L7¤p®Éªº¹ïJetBot¨t²Î¬[ºcªº­åªR¤Î¦b¥HNVIDIA Jetson NANO¬°®Ö¤ßªº´¼¯à¤p¨®¥­¥x¤Wªºµ{¦¡½s¼g¹ê§@¡A§¹¦¨¤p¨®±q°ò©óÂsÄý¾¹ªº»»±±¾Þ±±¡A¨ì¥Ø¼ÐÃѧO¦Û°ÊÁ×»Ù/¸òÂܪº¶µ¥Ø¹ê¬I¡A¾Ç­û¥i¥H¤F¸Ñ´¼¯à¾÷¾¹°ò¥»µ²ºc¤Î°ò©óµøıªº¤H¤u´¼¼z²`«×¾Ç²ß¤èªk¡A¦b¼Ò«¬°V½m/¼Ò«¬³¡¸p¤Wªº¹ê²{¬yµ{¡A¥]¬A°ò©óTensorRTªº¼Ò«¬Àu¤Æ¤Î³¡¸p¡C

  ½Òµ{¯S¦â
Ø   §¹¾ã©Ê¡G
1.   ¦bNVIDIA Jetson Nanoªº²`«×¾Ç²ß¡]DLI¡^­pºâ¤§¤W¡A¥[¤WAdafruitªº¾÷¹q±±¨îAPI¡A³z¹L¦h­Ó¹ê¥ÎªºµL¤H¨®¼ÒÀÀ®×¨Ò¡A¹ê»Ú¾Þ§@Data Collection¡BModel Training¡BApplication Deployment¤T¨BÆJ¡A§¹¾ã°õ¦æ”À³¥Î¶}µo“ªº¬yµ{¡C
2.   ¥»½Òµ{±Ð¾Ç³nÅ駹¾ã¦aµ²¦XNVIDIA Jetson¨t¦Cªº²`«×¾Ç²ßÀô¹Ò¡]¥]¬ACUDA¡BCUDNN¡BOpenCV¡BTensorRT¡BDeepStream¡^¡A¥H¤Î±`¥Îªº²`«×¾Ç²ß®Ø¬[¡]Caffe¡BTensorFlow¡BPytorch¡BKerasµ¥¡^¡A¨Ã¥B¾ã¦X°ªÀ³¥Î»ù­ÈªºDarknet-Yolo®Ø¬[»POpenPoseÅéºAÃѧO³nÅé¡C¦¹¥~¡A¥»±Ð¾Ç³nÅéÁÙ¥]¬AJetbot©Ò´£¨ÑªºCamera¡BGamePad¡BGPIOµ¥ÅX°Êµ{¦¡¤¶­±¡A§¡¥i¦bPythonÀô¹Ò¤¤»´ÃP½Õ¥Î¡C
 
Ø   ¹ê¥Î©Ê¡G
1.   ±qAIÃä½t­pºâªºTop-Downª¾ÃѹÏÃФJ¤â¡A±N¸H¤ù¤Æªºª¾ÃÑÂI«÷¦¨§¹¾ãªºÀ³¥Î
2.   ³z¹L¹ê»Ú¾Þ§@¡A¥Ñ²L¤J²`¡B¥h¿¾¦sµ×¦a´x´¤²Ó³¡ªº°õ¦æ§Þ¥©
3.   ©Ò¦³®×¨Ò§¡¯à¾A¦X”³q¥Î³õ´º“¡A¤£»Ý­n¦b¯S©wÀô¹Ò¤U¾Þ§@©ÎÀ³¥Î
4.   ©Ò¦³µ{§Ç¥N½X§¹¥þ¶}·½¡A«áÄò¥i¥H®Ú¾Úµø¨¤»Ý¨D¶i¦æ¥ô·N­×§ï
 
Ø   «K§Q©Ê¡G
1.   Jetbot¾Ç²ß¥­»O¥H³Ì¤Ö¼Æ¶qªº¹q¾¹¤¸¥ó¡]10­Ó¥H¤º¡^·f«Ø¦Ó¦¨¡A¥iÀH¨­Äâ±a
2.   ¦Û±a¹q¦À¨Ñ¹q¨t²Î»PµL½uWIFI³]³Æ¡A¥i¦b¤á¥~Àô¹Ò¡]µLºôµ¸¡^¤U°õ¦æ©Ò¦³À³¥Î
3.   ¥»½Òµ{±Ð¾Ç³nÅé¤w±N¦UºØ¾Ç²ß¤º®e¥þ³¡¿Ä¦X¡AµL»Ý§ó´«TF¥d
4.   ©Ò¦³±±¨î³£¥i¥H§Q¥Î¥ô¦ó¾Þ§@¨t²ÎªºPC/Notebook¡A³z¹LBrowser»»±±Jetbot
 
Ø   ©µ¦ù©Ê¡G
1.   AIÃä½t­pºâ³B©ó°_­¸¶¥¬q¡A¥¼¨Óµo®i¤´¦³«D±`¤jÅܤơA¦]¦¹§Ú­Ì爲¾Ç­û´£¨Ñ¤@­Ó²×¥Í§K¶OªºVIP§Þ³N¥æ¬y¸s¡A¦@¦P­±¹ïÀþ®§¦hÅܪº§Þ³N»P¥«³õ¡C
 
2.   Ãä½t¹Bºâ¡A¤TÁͶզ¨§Î¡C(1) On-Line Training §Ö³t±U°_¡AOn-Device Training ±N¦¨¬°Åã¾Ç¡B(2)¡uEdge AI ´¹¤ù+ AIºtºâªk¡v¬O¥þ²yÃä½t¹Bºâ§Þ³Nµo®iÁͶաB(3) ´Â¦VAI¾÷¾¹¾Ç²ß¬°¾É¦VªºÀ³¥Î¥ÍºA¨t¡Chttps://www.itri.org.tw/chi/Content/NewsLetter/contents.aspµL?SiteID=1&MmmID=5000&MSID=1036034731543070677
 
½Òµ{¹ï¶H
�� ¾A¦X¨ã¦³°ò¥»Python½sµ{¸gÅ窺¾Ç­û¤Î¶}µoªÌ
�� ¸Ë¸mMaker¡B¾÷¾¹¤HMaker¡B´O¤J¦¡¨t²Î¶}µoªÌ
�� ±q¨Æ¤H¤u´¼¼z¨t²Î³W¹º¡B«Ø¸m»PÀ³¥ÎªÌ
�� ±q¨Æ¤H¤u´¼¼zÃä½t¹Bºâ¸Ë¸m¡BAI¨t²ÎÀ³¥ÎªA°È¡B¨t²Î¾ã¦Xµ¥¤§¬ãµo¤H­û¤Î¥DºÞ¡C
�� ±q¨Æ¤H¤u´¼¼z³Ð·sÀ³¥Î¤§¥«³õ¬ã¨s¤Î­pµeºÞ²z¶}µo¸g²z¤H¡C
�� ¹ïAI ´¼¼zÀ³¥Î¹ê°È¬[ºc¤Î§Þ³N¡A¤ÎÀ³¥ÎªA°È·P¿³½ìªÌ
Á¿®v²¤¶
³¯ªa¿«Á`¸g²z
 
¡i¾Ç¾ú¡j
Ø   ­»´ä²z¤u¤j¾Ç«H®§ºÞ²z¬ã¨s©ÒºÓ¤h
Ø   °ê¥ß¥æ³q¤j¾Ç­pºâ¾÷¤uµ{¾Ç¨t¾Ç¤h
¡i¸g¾ú¡j
Ø   ¥xÆW/Ĭ¦{ ¦N®ú¨³¬ì§Þ¦³­­¤½¥qÁ`¸g²z
Ø   ¤¤¬ì°|°ª©Ê¯à¤¤¤ß°Æ¥D¥ô
¡i±M·~»â°ì¡j
Ø   ¥­¦æ­pºâ¡B´O¤J¦¡­pºâ¡B²`«×¾Ç²ßÀ³¥Î
Ø   NVIDIA DLI CUDA»{ÃÒÁ¿®v¡BNVIDIA DLI»{ÃÒ

½Òµ{¤jºõ

®É¶¡

½Òµ{¤jºõ

®É¼Æ

09:00

~

12:00

 

(§t¥ð®§)

Ø  ½Òµ{Á`Å餶²Ð¡G±qJetBot¾Ç²ß¥­»O¶i¤JÃä½t­pºâ»â°ì

Ø  JetBot¥­¥xµ²ºc¡A°t®MµwÅé«ü¼Ð¤ÎÀ³¥Î²¤¶¤Î¹ê§@

1.  ¥­¥xµ²ºc³]­p²z©À

2.  Jetson NANOªºµwÅé³W®æ¤¶²Ð

3.  Jetson NANO»P©PÃä³]³Æªº³s±µ¤è¦¡¡G40-Pin±µ¤f»¡©ú

4.  JetBot¾Ç²ß¥­»OµwÅé®i¶}¹Ï¡G©PÃä³]³Æ²¤¶

5.  ¹ê»ÚÀ³¥Î³õ´º©Ò»ÝªºMini Routerªº³]©w»P¨Ï¥Î

6.  û£°Ê»P¨Ï¥ÎJetBot¾Ç²ß¥­»O

¡m¹ê§@¡n³z¹LJupyterÀˬdJetBotªºIO³]³Æ¡G

- CSI Camera («DUSB camera)

¾÷¹q³]³Æ¡]JetBot«e¶i¡B«á°h¡BÂàÅs¡^

¡m¹ê§@¡n³z¹LJupyter¾Þ§@JetBot°õ¦æ²Õ¦X°Ê§@

¦Û¦æ­×§ï¥N½X¡A½Õ¾ã³t«×¡B¸ô½u¡B±±¨î­±ªOµ¥

 

Ø  Jetson Nano´£¨Ñ§¹¾ãªº²`«×¾Ç²ß¡]DeepLearning¡^¸ê·½Â²¤¶¤Î¹ê§@

n  ²`«×¾Ç²ß²­z¡G¯«¸gºôµ¸¡B¸ê®Æ¡B¼Ò«¬»P®Ø¬[ªºÃö¨t

n  Jetson Nanoªº²`«×¾Ç²ßSoftware Stack»¡©ú

¡m¹ê§@¡n°õ¦æYolo-Darknetû£°Ê¤£¦PCamera°µObject Detection

¡m¹ê§@¡n°õ¦æOpenPose«º¶Õ¡]ÅéºA¡^¿ëÃÑ

¡m¹ê§@¡nJetson Nano©Ê¯àÀu¤Æ¨BÆJ

¡m¹ê§@¡n­«·s°õ¦æYolo-Darknet»POpenPose©Ê¯à¤ñ¸û

ªþ¥ó¡BJetson Nano¶}µoÀô¹Ò¦w¸Ë¤Î¹Lµ{¤¤®e©ö¥X¿ùªº°ÝÃD¡]FAQ¡^

½T»{NANOÀô¹Ò·f«Ø¡G´ú¸Õ´X­Ó²`«×¾Ç²ßsamples

3

12:00

~

13:00

¤ÈÀ\

 

13:00

~

17:00

 

(§t¥ð®§)

Ø  JetBot´£¨Ñ§¹¾ãªº¾E²¾¾Ç²ß¡]Transfer Learning¡^¶}µo¸ê·½Â²¤¶¤Î¹ê§@

n  ¾E²¾¾Ç²ß²­z

n  JetBotªº¾E²¾¾Ç²ßSoftware Stack»¡©ú

¡m¹ê§@¡n³z¹LJupyter°õ¦æclassification

±Ä¶°¸ê®Æ -> °V½m¼Ò«¬ -> °õ¦æµ²ªG -> ¶i¦æ­×¥¿

¡m¹ê§@¡n©µ¦ù°Ñ¦Ò®×¨Ò¡A¦Û¦æ©w¸q»P¾Þ§@

¡m¹ê§@¡n¥Îclassification°õ¦æCollision Avoidance¡]Á×»Ù¡^¥\¯à

¡m¹ê§@¡n¥Îclassification°õ¦æFinding Path¡]´M¸ô¡^¥\¯à

n  Ãä½t­pºâÀ³¥Î¶}µo·sÁͶաGTraining vs Programming

- Deep/Transfer Learning»P¶Ç²ÎµøıÃѧOªº®t²§

¼vÅTLearning¿ëÃÑ«~½èªºÃöÁä¡GData Collectionªº«~½è

 

Ø  NVIDIA Inference Engine – TensorRT²¤¶¤Î¹ê§@

n  TensorRT ¥[³t inference ­ì²z

n  TensorRT ¤u§@¬yµ{

¡m¹ê§@¡nTensorRT¥[³tYoloªºObject Detection©Ê¯à

 

Ø  NVIDIA DeepStream 4.0µøÀW¤ÀªR³nÅ鲤¶¤Î¹ê§@

¡m¹ê§@¡n­×§ï config ÀÉ¡A§ïÅÜDeepStream 4.0¤u§@»P¿é¥X

¡m¹ê§@¡n²K¥[¼Ò«¬¡A§ïÅÜDeepStream 4.0ªº¥\¯à

4



 
 *¥D¿ì³æ¦ì«O¯d½Õ¾ã½Òµ{¤º®e¤§Åv§Q

½Òµ{¾Ç²ß¤¤¯A¤Îªº¶}µo¤u¨ã¡GJetpack¡AJupyter Notebook¡ATensorRT¡ATensorFlow¡AOpenCV¡APyTorch¡C­«ÂI±N¶°¤¤©óTensorRT¡C
¾Ç­û»Ý¦Û³Æµ§¹q©Î¥­ªO (¤Î¹q·½½u)
 
»ù®æ
1. ¤@¯ë¥Í¡G¨C¤H$18,000¤¸[³]³Æ+½Òµ{]¡A§tJetBot ´¼¯à¤p¨®®M¥ó¤@²Õ (¥i¦Û¦æÄâ¦^¨Ï¥Î)¡F
­Y¾Ç­û¦Û¦³³]³Æ¡A¥u­n½Òµ{¡A¨C¤H$9000¤¸[½Òµ{ Only]¡C
*³]³Æ¤º§t:
      • NVIDIA Jetson Nano Developer Kit *1®M
      • 32GB microSD card *1 ­Ó
      • 5V 6A DC¹q·½ *1®M
      • Jetson NANO±M¥ÎMIPI CSI Äá¹³ÀY *1­Ó
      • 3D¥´¦L¨®¤@®M(§t½ü¤l¡^ *1®M
      • TT¹q¾÷¨â¥u¤Î¹q¾÷ÅX°ÊªO *1®M
      • OLEDÅã¥Ü¹õ *1­Ó
      • WiFi USB Dongle *1­Ó
      • ¦æ°Ê¹q·½ *1­Ó
      • µL½u¸ô¥Ñ¾¹ *1­Ó
      • USB½u/¹q½u/Á³µ·µ¥½u§÷ *­Y¤z
  • ¦Û³Æ³]³Æ¤§ª`·N¨Æ¶µ:¬°½T«O³]³Æ§¹¾ã©Ê¤Î¬Û®e©Ê¡A½Ðª½±µ»PÂEÃ[°ê»Ú¦¶¸g²zÁpô¡A¥»½Òµ{¤£§t¾Ç­û¦Û¦³³]³Æ¤§°»¿ù¡C¦¶¸g²z: 02-2929-9388 #11¡AEmail: majestic@honghutech.com¡C
2.    [³]³Æ+½Òµ{]¦­³¾(12/13«e)Àu´f»ù¡G $16,000¡F [½Òµ{ Only]¦­³¾(12/13«e)Àu´f»ù: $8,000
3.    3¤H¥H¤W¹Î³ø¦A¨É¤E¤­§éÀu´f (¤u¬ã°|°£¥~)¡G
¶O¥Î
¦­³¾¹Î³ø»ù(12/13«e)
¹Î³ø»ù(12/14«á¡A«D¦­³¾)
³]³Æ+½Òµ{
$15,200
$17,000
½Òµ{ only
$7,600
$8,500
*¾Ç­û»Ý¦Û³Æµ§¹q©Î¥­ªO (¤Î¹q·½½u)*
* JetBot ¤p¨®¥¿±`¨Ï¥Î¤U¡A¥ÑÂEÃ[°ê»Ú´£¨ÑJetson Nano«O©T¤@¦~¤Î°â«áªA°È¡C
* ­Y±z¤w¦³³]³Æ¡A¥i¥H¥u¿ï”½Òµ{ Only”¡F¥t¥~­Yı±o¤@¥x¤£°÷¥Î¡A¥ç¥i¥H¹w¥[ÁÊ®M¥ó ($9,000)¡C
¶K¤ß´£¿ô
�� ³ø¦WºI¤î¤é´Á¡G2019/12/20[³]³Æ]¡A2019/12/20[½Òµ{ Only]¡C¦]±Ð«ÇªÅ¶¡¤Î±Ð¾Ç¡B³Æ®Æµ¥¦Ò¶q¡A­­30¦ì¾Ç­û¡A³ø¦W½Ð¦­¡C

ª`·N¨Æ¶µ
1. ¬°½T«O±zªº¤W½ÒÅv¯q¡A³ø¦W«á­Y¥¼¦¬¨ì¥ô¦ó¦^ÂСA·q½Ð¨Ó¹q¬¢¸ß¤è§¹¦¨³ø¦W¡C
2. ¦]¶·¥ý¹wÁÊJetBot ´¼¯à¤p¨®®M¥ó¡A¶}½Ò«e7¤Ñ(12/20«e)¡A»Ý¥ýú´Ú¡A¥BµLªk°h´Ú¡A¦ý¥i¥H§ó´«¥L¤H¡C­Y­ì³ø¦WªÌ¦]¬G¤£§J°Ñ¥[¡A¦ý±ý§ó´«¥L¤H°Ñ¥[¡A·q½Ð©ó¶}½Ò«e¤C¤é³qª¾¡C
3. ¾Ç­û»Ý¦Û±aµ§¹q (Win 10, Mac or Ubuntu)¤Î¹q·½½u
 ªþ¥ó

JetBot½Òµ{.DM-ªê§À-20191227-3.pdf

 


²¤¶


²£·~¾Ç°|½t°_
¡@¡@¨Ì¾Ú¦æ¬F°|¡u¬D¾Ô2008¡G°ê®aµo®i­«ÂI­pµe¡v¤U¤§¡u°ê»Ú³Ð·s¬ãµo°ò¦a¡v»P¡u²£·~°ª­È¤Æ¡v¨â­pµe¡A­º­«²£·~¬ì§Þ¤H¤~ªº®Ä¯à¡C

¡E911216¸g¬ì¦r²Ä09103373120¸¹¨ç¡G¸gÀÙ³¡±N¥»°|Äw³]¤u¬ã°|²£·~¾Ç°|¤§¤u§@¡A¦C¬°¦]À³²£·~µ²ºcÂ૬¡A´£