SeqSense shows how it built autonomous security robots on AWS
Japanese robotics startup SeqSense has built autonomous security robots using Amazon Web Services and Amazon’s platform for robotics
According to Japan’s national institute of population and social security research, the working population in Japan is expected to decrease by 30 percent in the next 30 years. A rapidly aging population in Japan causes a shortage of labor in the security space.
SeqSense, a robotics startup company in Japan, aims to solve this challenge by providing autonomous security robot service. They developed an intelligent security robot SQ-2. that patrols inside buildings.
It navigates by itself using pre-created map of a floor, patrol multiple floors by communicating with elevators. Their robot has been deployed in more than 30 office buildings, shopping malls, and airports in Japan.
Their core technologies are localization, navigation, and mapping using super wide field-of-view lidar. They developed their original 3D lidar sensor for 3D mapping and localization. It enables the robot to navigate in narrow aisle or crowded environment.
The robot also has a high resolution camera, 360-degrees camera with 3 fisheye lenses, microphones, and speakers. Operators can monitor the robot remotely and communicate through the robot. The robot moves around up to 6 hours with 1.5 hour charge and go back to the charging station by itself.
Using autonomous robots for security has several advantages. For example, surveillance cameras are often used for building security. However, surveillance cameras sometimes have blind spots, and setting up additional security cameras takes cost and time.
In contrast, the SQ-2 robot patrols the whole building by taking lift and security guards can choose routes that reduce or eliminate blind spots. In addition, the existence of a physical robot is also useful to deter crimes like shoplifting without having human security guards.
In the following video, you’ll learn how SeqSense utilizes AWS for capabilities like connectivity, remote operation, and fleet monitoring. This post dives deep into their architecture and challenges they solve with AWS services.
Challenges
Developing an autonomous robot is a complex and time-consuming task, requiring a wide variety of technologies from device to cloud. For example, securely operating and monitoring robot fleets requires collecting telemetry and video data for real-time dashboards.
Robot developers can improve the robot software by analyzing the collected data and provide OTA update through cloud.
As a startup company, SeqSense wants to focus on value-differentiation and its core competencies like 3D LiDAR and navigation.
However, cloud platform is also important for their business because their robot is designed to be operated by human security guards through a web console. They have to develop and operate the cloud platform as well as the robots with limited number of engineers.
Solution and architecture overview
Their robot is always connected to the internet via 4G cellular connectivity and sending telemetry and video to the cloud. However, it can operate completely by itself even when it is disconnected.
Security guards and operators monitor and manage robots from their web console. It shows the status of robot, map, position and video of a robot. Users command a robot to patrol specific waypoints or go back to charging station.
SeqSense wanted fully managed services that help reduce the cost of overseeing infrastructure. They leverage services like AWS IoT Core and Amazon Kinesis Video Streams for ingesting robot data. SeqSense develops and manages all of their cloud infrastructure with only 4 engineers.
Collecting telemetry
They use AWS IoT Core to collect telemetry from the robot. IoT Core lets devices easily and securely connect to cloud and interact with cloud applications. The data collected from the robot will be stored in multiple database services depending on the use case.
For example, patrol results are stored in Amazon Relational Database Service (Amazon RDS), the latest robot statuses like operation mode, errors and battery percentage are stored in Amazon ElastiCache for Redis, and the historical data of robot position is stored in Amazon TimeStream.
Their cloud connectivity software on the edge are written in Golang. They have developed a library to use AWS IoT Core features from Golang and published it as an open source software on GitHub.
Remote operation
IoT Core is also used for remote operation. Security guards or building facility owners create a patrol route in their web console and set the route for each robot. A robot usually patrol the building along the specified route.
When it found an obstacle on the way to waypoint, it avoid the obstacle. However, if it fails to detour and can’t reach to the waypoint, it notifies to its owner. The owner can skip a waypoint, command the robot to stay, and go back to the charging station.
Video streaming and recording
SQ-2 has four cameras and users can see live and on-demand video from the web console. The video streamed to the cloud is securely stored into Amazon Kinesis Video Streams for on-demand playback.
Users can easily download or playback the media of a certain period. SeqSense also uses the stored media to improve the robot software. They have developed Kinesis Video Streams Producer and Consumer Library for Golang.
Shortage of labor
SeqSense aims to solve a shortage of labor in the security space by developing intelligent security robots. By using AWS services, they can make their service reliable, durable, and easy to maintain, scale, and extend.
Especially, AWS IoT Core and Amazon Kinesis Video Streams help them easily develop and operate their core features like autonomous patrol and remote monitoring.