Patterns for cloud-enabled robots
summary
Shows concrete patterns using Greengrass V2 components to push modular ROS or Docker apps to remote robots, perform staged rollouts (start at 5% of fleet), stop at configurable failure thresholds (example stop at 20%), and auto-rollback; demonstrates edge ML workflow using SageMaker Ground Truth for labeling, SageMaker training, Neo for edge optimization, then deploying models to Spot via Lambda components (object detection of fire extinguishers that triggers SNS alerts); contrasts low-latency teleoperation with Kinesis Video Streams with WebRTC and higher-latency video archival to KVS.
AI-written summary · about
chapters (4)
- 0:00Start
- 1:25Common challenges while building smart robots
- 28:38Video recording at the edge for robots
- 34:53Example: Deploy and run ROS2 Foxy Docker containers
related talks
28mBuild smart robots and operate them easily with AWS Robotics
33mFeeding the Cloud: Designing Data-Driven Robotics Systems
48mBuilding Robotics Applications at Scale using Open Source from Zero to Hero
25mBuilding a PaaS for Robotics with Kubernetes
14mReliability Engineering for Hybrid Robot-Cloud Systems
21mBuilding a PaaS for Robotics with Kubernetes
01h 00mTeaching a robot new tricks
28mTerraform, Infrastructure as Code, and the Robot Dog