Cloud robotics allows you to deploy robotics applications using cloud computing, cloud storage, communication, and other cloud technologies. By storing the decision-making component (“brain”) in the cloud, the robots can be low-cost and lightweight with only low-level controls on-board. The robot can use the brain to access large databases with task planners, environment models, processing and communication capabilities, and more.
Cloud robotics enables robots to access:
- Knowledge bases such as updated libraries of images and maps
- Parallel computing on demand for statistical modeling and analysis, motion and task planning, and robot learning
- Communication resources for robots to share information such as trajectories, control policies, and outcomes
- A development environment for engineers including simulation platforms, analysis and visualization tools, training datasets, and benchmarks
Cloud robotics is being used in self-driving cars, medical robots, personal and industrial robots, and other autonomous applications. A self-driving car with cloud capabilities can access maps and environment models and integrate them with live-streamed sensor data to precisely locate its position and avoid collisions. The data obtained on-the-go can also be shared with other connected cars via wireless cloud connectivity.
MathWorks Cloud provides instant access to MATLAB® and other products and services. With MATLAB Online™, you can quickly run your algorithms and verify the results in a web browser. You can use MATLAB® on virtual machines in Amazon Web Services (AWS), Microsoft Azure®, and public cloud environments and take the advantage of high-performance GPUs.
Using MATLAB Parallel Server™, you can run MATLAB® programs and Simulink® simulations on clouds and clusters such as Amazon EC2. You can also integrate Simulink with third-party visualization platforms running on the cloud to run your algorithms in a photorealistic 3D environment. For example, Simulink® can be integrated with Unreal Engine® running on the cloud. You can use MathWorks published reference architectures for a head start.