Cloud Robotics Study

ABI Research Report: Cloud Robotics

  1. Why cloud robotics?

    Developers can offload the computational heavy tasks to cloud. As robot systems scale, this becomes extremely important.

    Companies turn to cloud-based services to expand their offering in value-added services.

  2. Robotics systems key functions

robotics technology summary

  1. Services vs. Cloud

    services_vs_cloud

    1. Most robotics companies with cloud-robotics projects are working closely with military.

A Survey of Research on Cloud Robotics and Automation

Cloud Robotics: Architecture, Challenges and Applications

Computational challenges; communication challenges, optimization challenges, security challenges

Computation Sharing in Distributed Robotic Systems: A Case Study on SLAM

  1. Take GMapping algorithm as an example: 98% of the calculation is on mapping the identified particles to the existing data.
  2. The idea is relatively simple: send the particles information to the remote system for a matching calculation, and receive the result. Then check if a re-sampling is needed and take further actions. (ask for missing data, etc.)
  3. Clearly, the information is in cleartext. This gives space for the researchers to work on privacy-preserved SLAM.

Privacy in MRS and Cloud Robotics

Coordinated multi-robot planning while preserving individual privacy

This paper discusses the scenario where two robots work in the same environment, and their path may have overlapping which potentially causes collision.

The concept of Secure Multi-Party Computation (SMC)

My thoughts

  1. SLAM is the key difference between robotics systems and other IoT systems.
  2. Identify those functions that requires real-time processing. While research works like “A Cloud-Based Visual SLAM Framework for Low-Cost Agents” addresses the usage of cloud in SLAM of robots, such implementation is still not accepted by the robotics community.
  3. Cloud-robotics industry will grow dramatically in the future. However, the usage of cloud is not for the in-built functions of the robots, but the extensive add-on services provided by the robots.
  4. Encryption over messages transmitted between robots are effectively useful. However, it does not have many real-world usages.
  5. One of the possible directions is to build privacy-preserved MRS SLAM protocol.