Skip to content

Examples

Jump to bottom
John edited this page Jan 16, 2020 · 5 revisions

This page lists the examples provided with JetBot.

Make sure your robot is connected to WiFi as described in the software setup

Example 1 - Basic Motion

In this example we'll control JetBot by programming from a web browser.

  1. Connect to your robot by navigating to http://<jetbot_ip_address>:8888

  2. Sign in with the default password jetbot

  3. Navigate to ~/Notebooks/basic_motion/

  4. Open and follow the basic_motion.ipynb notebook

    Make sure JetBot has enough space to move around.

Example 2 - Teleoperation

This example requires a gamepad controller connected to your workstation.

In this example we'll drive JetBot remotely, view live streaming video, and save snapshots!

  1. Connect to your robot by navigating to http://<jetbot_ip_address>:8888

  2. Sign in with the default password jetbot

  3. Shutdown all other running notebooks by selecting Kernel -> Shutdown All Kernels...

  4. Navigate to ~/Notebooks/teleoperation/

  5. Open and follow the teleoperation.ipynb notebook

Example 3 - Collision avoidance

In this example we'll collect an image classification dataset that will be used to help keep JetBot safe! We'll teach JetBot to detect two scenarios free and blocked. We'll use this AI classifier to prevent JetBot from entering dangerous territory.

Step 1 - Collect data on JetBot

We provide a pre-trained model so you can skip to step 3 if desired. This model was trained on a limited dataset using the Raspberry Pi V2 Camera with wide angle attachment.

  1. Connect to your robot by navigating to http://<jetbot_ip_address>:8888

  2. Sign in with the default password jetbot

  3. Shutdown all other running notebooks by selecting Kernel -> Shutdown All Kernels...

  4. Navigate to ~/Notebooks/collision_avoidance/

  5. Open and follow the data_collection.ipynb notebook

Step 2 - Train neural network

Option 1 - Train on Jetson nano

  1. Shutdown your robot and remove the micro USB power cable.

  2. Power the Jetson Nano by using the 5V wall power supply.

  3. Connect to your robot by navigating to http://<jetbot_ip_address>:8888

  4. Sign in with the default password jetbot

  5. In the Jupyter Lab tab, navigate to ~/collision_avoidance

  6. Upload the collision avoidance training notebook to this folder

  7. Open and follow the train_model.ipynb notebook

Option 2 - Train on other GPU machine

  1. Connect to a GPU machine with PyTorch installed and a Jupyter Lab server running

  2. Upload the collision avoidance training notebook to this machine

  3. Open and follow the train_model.ipynb notebook

Step 3 - Run live demo on JetBot

  1. Power your robot from the USB battery pack

  2. Connect back to your robot by navigating to http://<jetbot_ip_address>:8888

  3. Sign in with the default password jetbot

  4. Shutdown all other running notebooks by selecting Kernel -> Shutdown All Kernels...

  5. Navigate to ~/Notebooks/collision_avoidance

  6. Open and follow the live_demo.ipynb notebook

    Start cautious and give JetBot enough space to move around.

Video

This video shows multiple JetBots running collision avoidance

Example 4 - Object Following

In this example we'll have JetBot follow an object using a pre-trained model capable of detecting common objects likePerson, Cup, and Dog. While doing this, JetBot will run the collision avoidance model from Example 3 to make sure it stays safe!

  1. Download the object detection model according to the table below

    JetBot SD Card Version Model
    v0.3 ssd_mobilenet_v2_coco.engine
    v0.4 (latest) ssd_mobilenet_v2_coco.engine

  2. Connect to your robot by navigating to http://<jetbot_ip_address>:8888

  3. Shutdown all other running notebooks by selecting Kernel -> Shutdown All Kernels...

  4. Navigate to ~/Notebooks/object_following/

  5. Upload the pre-trained ssd_mobilenet_v2_coco.engine model to this folder

    Also make sure the collision avoidance model from Example 3 is in ~/Notebooks/collision_avoidance

  6. Open and follow the live_demo.ipynb notebook

    Start cautious and give JetBot enough space to move around.

Video

This video shows JetBot following a person and avoiding obstacles

Next

Make JetBot smarter

  • Collect more collision avoidance data
  • Try out different neural network architectures (the torchvision package has lots!)
  • Modify the collision avoidance example for a new task (ie: cat / no cat. if cat then run)

Create something entirely new!

  • Modify the collision avoidance example for your own project
  • Try out some new hardware with Jetson Nano. It's easy with Jetson GPIO and Adafruit Blinka

Share it with us

Analytics

Clone this wiki locally