Postmates, the on-demand delivery company that operates across the United States presented a new all-electric autonomous delivery robot equipped with the latest…
Overview
Postmates unveiled its new all-electric autonomous delivery robot named Serve, equipped with the NVIDIA Jetson AGX Xavier module, at GTC Silicon Valley. The robot is designed to efficiently transport goods over short distances while respecting urban environments.
What You'll Learn
1
How to leverage NVIDIA Jetson AGX Xavier for autonomous robotics applications
2
Why using TensorFlow and TensorRT enhances AI model performance in robotics
3
How to implement real-time obstacle detection using Velodyne LIDAR sensors
Prerequisites & Requirements
- Understanding of AI/ML concepts and robotics
- Familiarity with TensorFlow and TensorRT(optional)
Key Questions Answered
What capabilities does the Serve delivery robot have?
The Serve delivery robot can carry up to 50 pounds of goods and has a range of 30 miles. It operates at walking speed and is designed to work alongside the Postmates fleet for efficient short-distance deliveries.
How does the Serve robot navigate urban environments safely?
The Serve robot uses GPUs for safe navigation in city environments. It employs Velodyne LIDAR sensors to create a real-time virtual picture of the surroundings, ensuring it can move autonomously on sidewalks and in neighborhoods.
What technology underpins the Serve robot's AI capabilities?
The Serve robot is powered by the NVIDIA Jetson AGX Xavier module, which enables the deployment of end-to-end AI robotics applications. The system utilizes TensorFlow for model training, which is then converted to TensorRT for deployment.
What is the expected launch area for the Serve robot?
Postmates plans to launch the Serve robot in the Los Angeles area before expanding to additional cities across the United States, aiming to provide zero-cost deliveries to consumers.
Key Statistics & Figures
Payload capacity
50 pounds
The maximum weight the Serve robot can carry.
Operational range
30 miles
The distance the Serve robot can travel on a single charge.
Speed improvement
40x speedup
Achieved by using FP16 half-precision with NVDLA on the Xavier platform.
Technologies & Tools
Some links below are affiliate links. We may earn a commission if you make a purchase.
Hardware
Nvidia Jetson Agx Xavier
Used for AI processing in the Serve robot.
Sensors
Velodyne Lidar
Provides real-time environmental mapping for navigation.
Software
Tensorflow
Used for model training in the robot's AI system.
Software
Tensorrt
Used for optimizing and deploying AI models on the Xavier platform.
Key Actionable Insights
1Integrating NVIDIA Jetson AGX Xavier into robotics projects can significantly enhance AI processing capabilities.Utilizing this technology allows for the development of advanced autonomous systems that can operate efficiently in complex environments.
2Employing Velodyne LIDAR sensors is crucial for real-time environmental mapping in autonomous robots.This technology enables robots to navigate safely by detecting obstacles and understanding their surroundings, which is essential for urban delivery applications.
3Using TensorFlow for model training and TensorRT for deployment can lead to significant performance improvements.The Serve robot achieved over 40x speedup by converting models to TensorRT, showcasing the importance of optimizing AI workflows for real-time applications.
Common Pitfalls
1
Failing to optimize AI models for real-time performance can lead to inefficient robot operations.
Without proper optimization techniques like those provided by TensorRT, robots may struggle to process data quickly enough to navigate safely.