Satinder Singh won the Jetson Project of the Month for DeepWay, an AI-based navigation assistance system for the visually impaired. The project…
Overview
The article highlights the DeepWay project, an AI-based navigation aid designed for visually impaired individuals, developed by Satinder Singh. Utilizing the NVIDIA Jetson Nano Developer Kit, the system offers guidance through tactile feedback and voice prompts, enhancing mobility and safety for users.
What You'll Learn
1
How to create an AI-based navigation aid for visually impaired users
2
Why using tactile feedback enhances navigation safety for visually impaired individuals
3
How to train a convolutional neural network using PyTorch and Keras
4
When to choose the U-Net model for semantic segmentation tasks
Prerequisites & Requirements
- Basic understanding of AI and machine learning concepts
- Familiarity with NVIDIA Jetson Nano and Arduino Nano(optional)
- Experience with Python programming and neural network training
Key Questions Answered
How does the DeepWay project assist visually impaired individuals?
The DeepWay project assists visually impaired individuals by providing navigation guidance through tactile feedback and voice prompts. It uses an AI system running on the NVIDIA Jetson Nano to monitor the user's path and alert them to obstacles, enhancing their mobility and safety.
What technologies are used in the DeepWay project?
The DeepWay project utilizes the NVIDIA Jetson Nano Developer Kit, Arduino Nano, a webcam, and a USB audio adapter. These components work together to create a portable and affordable navigation aid for visually impaired users.
What is the training process for the convolutional neural network in DeepWay?
Satinder collected 10,000 images across three classes to train his convolutional neural network. He used a GPU-enabled virtual machine on Microsoft Azure and employed both PyTorch and Keras, ultimately selecting the U-Net model trained in Keras for its superior performance.
What is the estimated cost of the DeepWay navigation aid?
The estimated cost of the DeepWay navigation aid is less than $200 USD, making it an affordable option for visually impaired individuals seeking assistance in navigation.
Key Statistics & Figures
Number of images collected for training
10,000
Satinder collected these images across three classes for left, center, and right positions within a lane.
Estimated cost of the DeepWay navigation aid
less than $200 USD
This affordability makes it accessible for many visually impaired individuals.
Number of classes for model training
3
The classes represent left, center, and right positions within a lane.
Technologies & Tools
Some links below are affiliate links. We may earn a commission if you make a purchase.
Hardware
Nvidia Jetson Nano Developer Kit
Used as the main processing unit for the AI-based navigation system.
Hardware
Arduino Nano
Controls the servo motors that provide tactile feedback to the user.
Software
Pytorch
Used for training the convolutional neural network.
Software
Keras
Also used for training the convolutional neural network, with the U-Net model selected for its performance.
Cloud
Microsoft Azure
Provided GPU-enabled virtual machines for model training.
Key Actionable Insights
1Consider integrating tactile feedback mechanisms in assistive technologies to enhance user experience.Using tactile feedback not only aids in navigation but also allows users to maintain their auditory senses, which are crucial for safety in everyday environments.
2Leverage cloud computing resources for training complex AI models efficiently.By utilizing GPU-enabled virtual machines, developers can significantly speed up the training process for neural networks, allowing for quicker iterations and improvements.
3Explore open-source projects like DeepWay to foster innovation in assistive technologies.Open-source projects provide valuable insights and frameworks that can inspire new solutions for social good, particularly in areas like accessibility.
Common Pitfalls
1
Neglecting the importance of user feedback in assistive technology development.
Failing to incorporate user feedback can lead to solutions that do not effectively meet the needs of visually impaired individuals, reducing the project's impact and usability.
Related Concepts
Ai-based Navigation Systems
Assistive Technology For Visually Impaired
Convolutional Neural Networks
Open-source Projects In AI