NVIDIA SMART CAR

We made this project when we participated Hack x SJTU compotition.
Thanks NVIDIA for their technical support

Group member:

Jin Yongxu
Luo Anqi
Gou Minghao
Wang Tao
Qian Xinxin

Function Detail

We implemented a smart car using Arduino and TX1. The smart car do some movement as following:

• Move forward when the camera detects a car
• Move Back when it detects a chair
• Turn left when it detects a cat
• Turn right when it detects a person
• Turn left then move forward when it detects a bird
• Turn right then move forward when it detects a bicycle
• Stop immediatily when the car meets an obstacle(using ultrasonic sensor)

Techniques

• Caffe Deeplearning framework
• Arduino
• ROS
• OpenCV

Picture

Files

• car4.c is (controlling arduino)
• main.cpp is (controlling TX1)

Deployment

You should use ROS to create a function package named ros_caffe
Then create a node named ros_caffe_test

Run

  catkin_make
  roscore
  rosrun rosserial_python serial_node.py /dev/ttyACM0
  rosrun ros_caffe ros_caffe_test

Here is the official website of Hackthon.