This robot not only looks like a dog but he learns like one too! Researchers from Florida Atlantic University’s Machine Perception and Cognitive Robotics…
Overview
The article discusses Astro, a robotic dog developed by researchers at Florida Atlantic University, which learns through a deep neural network rather than preprogrammed code. Astro is designed to assist the visually impaired and medical professionals, utilizing advanced sensors and NVIDIA technology for processing sensory inputs.
What You'll Learn
1
How to train a robotic system using deep neural networks
2
Why using multiple sensors enhances robotic capabilities
3
How to utilize NVIDIA TX2 Modules for processing sensory inputs
Prerequisites & Requirements
- Understanding of deep learning concepts and neural networks
- Familiarity with NVIDIA TX2 Modules and their applications(optional)
Key Questions Answered
What capabilities does Astro the robotic dog have?
Astro is equipped with over a dozen sensors, including cameras, radar, and a directional microphone, allowing it to process environmental inputs across multiple modalities. It can currently respond to commands like 'sit,' 'stand,' and 'lie down,' and aims to understand hand signals and perform facial recognition in the future.
How is Astro trained to perform tasks?
Astro is trained using inputs to a deep neural network, allowing it to learn from experience rather than relying on preprogrammed code. The training process involves using NVIDIA GeForce RTX 2080Ti GPUs with frameworks like PyTorch and TensorFlow for action classification and model compression.
What are the main applications of Astro?
Astro is designed to assist the visually impaired as a service dog and to support medical professionals with diagnostic monitoring. Its capabilities are aimed at benefiting humanity through various tasks that mimic those of a traditional service dog.
Key Statistics & Figures
Computational capability
up to four trillion computations a second
This performance is achieved through the onboard GPUs of the NVIDIA TX2 Modules.
Weight of Astro
60 pounds
This weight contributes to its stability and mobility as a robotic service dog.
Technologies & Tools
Some links below are affiliate links. We may earn a commission if you make a purchase.
Hardware
Nvidia Tx2 Modules
Used for processing sensory inputs and performing computations.
Hardware
Nvidia Geforce Rtx 2080ti
Utilized for training the action classification models.
Software
Pytorch
Framework used for training the action classification models.
Software
Tensorflow
Used for model compression in the training process.
Key Actionable Insights
1Implementing deep learning training for robotic systems can significantly enhance their adaptability and functionality.By utilizing deep neural networks, developers can create robots that learn from their environment, making them more effective in real-world applications.
2Incorporating multiple sensor modalities can improve a robot's ability to interact with its environment.Astro's use of various sensors allows it to gather comprehensive data, which is crucial for tasks such as navigation and assistance.