Presentation is loading. Please wait.

Presentation is loading. Please wait.

Ben Schaeffer Gehad Shaat Jessie Truong Nathen Upperman Tin Nguyen Portland State University - Winter 2011.

Similar presentations


Presentation on theme: "Ben Schaeffer Gehad Shaat Jessie Truong Nathen Upperman Tin Nguyen Portland State University - Winter 2011."— Presentation transcript:

1 Ben Schaeffer Gehad Shaat Jessie Truong Nathen Upperman Tin Nguyen Portland State University - Winter 2011

2  Introduction to Einstein  What we did and what we used  Image Processing  Software management Portland State University - Winter 2011

3 BeforeAfter

4 Portland State University - Winter 2011

5  Bluetooth API and implementation  Servo controller application  Android application (robot remote control) with voice recognition  object and human detection application  Text-to-speech  Integration of all previously mentioned component into einsteinBrain  Efficient Power system  CUDA & TBB base program Portland State University - Winter 2011

6  IDE ◦ Visual studio 2008 (developing C++ application) ◦ RobotC (developing NXT application) ◦ Appinventor ( developing android application)  Libraries ◦ OpenCv 2.2 C/C++ ◦.NET C++ Framework ◦ CUDA toolkit ◦ NVIDIA Performance Primitives (NPP) library ◦ GPU Computing SDK ◦ TBB (Threading Building Blocks)  Hardware ◦ NXT ◦ Maestro servo controller ◦ DC motors with Hitechnic controller ◦ self-built power distribution system ◦ 16 V battery  Subversion for managing resources and documents Portland State University - Winter 2011

7  Use neural network to improve Einstein's Artificial intelligence  Make him better good looking!  replace NXT with a powerful microcontroller  more balanced weight of the body  improve power efficiency  add wireless camera  add more sensors & stereo camera  replace the neck motor with servo  use of CUDA & TBB to improve speed Portland State University - Winter 2011

8  We did the following processing on images ◦ Find colored objects ◦ Detect how many beats per minutes for a drumstick ◦ Face detection with area estimation Portland State University - Winter 2011

9

10

11

12  Load the picture normally  Upload to gpu  Process image using GPU functions src = imread("file.jpg", CV_LOAD_IMAGE_GRAYSCALE); GpuSrc.upload(src); cv::gpu::matchTemplate(GpuSrc,GpuTmplt,GpuDst, CV_TM_CCORR_NORMED); Portland State University - Winter 2011

13

14 Power solutions are critical. Camera choice is important. Organize all SW components early. Design for speed!


Download ppt "Ben Schaeffer Gehad Shaat Jessie Truong Nathen Upperman Tin Nguyen Portland State University - Winter 2011."

Similar presentations


Ads by Google