The AI Podcast avatar

Dexa/The AI Podcast

Learn more
Published Oct 21, 2019

Clean Sweep: Tokyo Robotics Company Builds Tidying Robots - Ep. 101

Explore how Preferred Networks is revolutionizing the robotics industry with tidying robots that use cutting-edge computer vision and natural language processing. Jun Hatori delves into the synergy between hardware and software, the challenges of data collection, and the future of autonomous cleaning technology.
Episode Highlights
The AI Podcast logo

Topics covered

Popular Clips

Episode Highlights

  • Tidying Robot

    Preferred Networks has developed a personal robot capable of autonomously tidying rooms using advanced computer vision and natural language processing (NLP). explains that the robot can recognize items in a room, create a task list, and execute tasks like picking up objects and placing them in designated spots 1. This innovation goes beyond simple vacuuming robots by integrating human-robot interaction technologies, allowing the robot to understand and respond to spoken commands and gestures 2.

    We help build a system that can work on the real world task automatically and also can communicate with people in an intuitive way.

    ---

    The robot's ability to dynamically adjust its actions based on user input makes it a versatile tool for home organization.

       

    Object Detection

    The robot's object detection capabilities are crucial for its tidying function. describes how the robot uses advanced computer vision to recognize and classify various items, such as clothing, trash, and stationery, and then moves them to appropriate locations 3. This process, known as grasp planning, involves the robot determining the best way to pick up and handle each object based on its shape and placement.

    Recent advances in the computer vision community have finally enabled building such robust computer vision system for hundreds of objects.

    ---

    These advancements have made it possible for the robot to operate effectively in diverse real-world environments.

       

    Gesture Recognition

    Gesture recognition technology allows the robot to adjust its actions based on human commands. explains that the robot can dynamically change its task planning when given spoken or gesture commands by the user 1. For example, if the robot is about to place a sock in the laundry bag, a user can interrupt and instruct it to throw the sock away instead.

    The robot system can dynamically update its task planning result and move it to the trash bin instead.

    ---

    This flexibility enhances the robot's usability, making it a more effective tool for home tidying.

Related Episodes