COLUMNS Advanced Technology Megatrends Robotics
  • URLをコピー Copy
Robotics Part 2
A world where we work, learn, play, relax, and eat with robots every day

Notable Robotic Technologies

Yoshihiro Mori, Senior Research Fellow
Yuta Kakimoto, Future Vision Center
2 April 2020

Technologies Required by Service Robots: Wide-ranging & still under development

A robot can be described basically as technology that carries out the following functions: conversion of a displacement registered by an input sensor to an electrical signal, entry of this signal into its system, and the physical movement of an output device such as an actuator following the system’s internal control logic.

Since general industrial robots are not expected to operate with the safety constraints that come with having humans in close proximity, the interplay of input, output, and control can be achieved with relatively unsophisticated technology. However, specified industrial robots and service robots that work in full collaboration with humans require a wider range of technologies. This report uses service robots to discuss several technologies that can be applied to the input, output, and control of robots (Figure 1).

Figure 1

Technologies Used with Robots

Source: Mitsubishi Research Institute, inc.

Robots employ technologies spanning a wide range of fields. While robotics is often categorized as simply one field within engineering, it combines several fields including mechanics, information technology, telecommunications, and chemistry. For example, if a robot is to work alongside humans, the distance between the humans and the robot needs to be understood accurately. Consideration must also be given to its control and materials used in its design to prevent harm to humans even if they were to come into contact with the robot. Service robots that meet these requirements are already used in various settings. While some of the technologies used for service robots are currently in use, others still require further development before they can work alongside humans. Some of the technologies that are expected to progress rapidly in the near future are highlighted below.

Notable Trends in Robotic Technologies


Integrated Tactile Sensors

The tactile sensor is equivalent to the sense of touch, one of the five human senses. When touching an object, humans perceive how strongly they are touching an object and whether the object will slip when lifted. However, current tactile sensors only have a single function based on simple force detection, and one device alone cannot measure subtle changes in force.

However, research on sensor integration is being conducted from a variety of perspectives. For example, the Advanced Telecommunications Research Institute International (ATR) has been studying the development of integrated contact sensors that apply microelectromechanical system (MEMS) technology. Figure 2 provides a visualization of the structure behind the integrated multi-axis tactile sensor, a type of integrated contact sensor. By measuring pressure and shear force at the same time, the robot can be used to hold objects with optimal force.

Figure 2

Basic Structure of Integrated Multi-Axis Tactile Sensor

Note: This technology is being developed by Ritsumeikan University and Niigata University, led by Haruo Noma, Professor at Ritsumeikan University.

Source: Intelligent Robotics and Communication Laboratories “MEMS Tactile Sensor” Intelligent Robotics and Communication Laboratories
https://irc.atr.jp/en/research-projects/human_beh_ana/mems-tactile-sensor/ (Accessed: January 29, 2020)

BMI & Myoelectric Sensor

In addition to the autonomous control of robots, other technologies that refine the control of robots are under development. Control technologies that differ significantly from conventional ones include brain-machine interface (BMI), based on the measurement and analysis of brain activity, and myoelectric sensors, which measure and analyze the electric signals generated by muscular movement. These input systems are characterized by their use of human biosignals and the innovative use of human intentions that precede motion and decision-making.

Biosignal measurement and signal processing are the technologies that lie at the core of BMIs and myoelectric sensors. BMIs use biosignals such as brain waves, brain magnetic fields, and cerebral blood flow while myoelectric sensors use myoelectric potential. However, biosignals are extremely weak, and in the measurement of brain waves, for example, a shield box is required to eliminate external noise in electroencephalogram measurements. Thus, BMI is still limited in its use. Nevertheless, measurement in typical situations and with portable devices is starting to become possible with improvements in measurement instruments like electrodes and with progress in pattern recognition and state estimation using machine learning. In turn, such advancements will make possible easy measurement and high-precision analysis of complex data regardless of location. BMI and myoelectric sensors will begin to surface as marketable products.

As technology to control equipment using human intentions, BMI and myoelectric sensors have been considered for use by people with physical disabilities to move prosthetic hands or wheelchairs. Recently, technologies are emerging that enhance physical capabilities instead of only replacing or assisting such functions. One example can be found in an experiment on the use of BMI sensors to allow a subject to manipulate a robotic arm utilizing brain waves while at the same time using both of the subject’s own arms (Figure 3).

Figure 3

Manipulating robot arm using brain waves through BMI while using both arms

Source: (2018) “First third arm than can be manipulated only by thinking” Japan Science and Technology Agency (JST)
https://www.jst.go.jp/pr/announce/20180726/index.html (Accessed: January 29, 2020)

Expectations are growing for BMIs employed not only as a motor output function but as a sensory input function. One concrete research project in progress is developing BMI to induce sensory consciousness by receiving and processing external information, instead of sensory organs such as an eye or an ear, and transmitting signals to the sensory area of the brain (Figure 4). Connecting the brain and the machine for input rather than solely for output prominently indicates that the link between humans and robots has progressed to the phase of seamless fusion. This gives hope for the establishment of high-speed, accurate, and large-scale processing of sensory information that is difficult via the human body. However, further discussion from an ethical perspective is necessary before such technological advances.

Figure 4

Example of Artificial Retina System

Source: Nidek Co., Ltd. “Types of Visual Prostheses” Nidek Co., Ltd. Website
https://www.nidek-intl.com/aboutus/artificial_sight/about_artificial_sight/type.html (Accessed: January 29, 2020)

AI & Cloud Robotics

The application of artificial intelligence (AI) to robot control has already begun with industrial robots. Although control must first be improved, since service robots require a greater degree of refinement in their control than industrial robots they have an even greater need for AI.

AI will first be applied to the control of robots that operate independently. However, as control too becomes more sophisticated, the processing capacity of individual robots is expected to reach its limit even with the combined use of chips dedicated to AI. In the future, it is expected that a shift will occur to capitalize on the CPU power of cloud computing. Control achieved through the combination of the cloud and a chip will allow the coordinated control of multiple robots. As a result, it will become possible to collectively process large volumes of data obtained from multiple robots with AI. Such processing capabilities are expected to enable the optimized control of robots as a whole rather than the status quo in which data obtained from a single robot can only be used for individual optimization.

Once the cloud becomes able to control a large number of robots, processing delays may become a challenge. In such a case, edge computing will likely be applied between the robots and the cloud computer. Data processing can be accelerated by having edge computing respond to data which does not need to be uploaded to the cloud and urgent issues. Also, the overall response of the system can be improved by having edge computing also handle the preprocessing of data to be uploaded to the cloud computer as well.

As a rule, cloud robotics will feature the wireless connection of autonomously operating robots and edge-cloud computers. However, wireless technologies by nature cannot guarantee the quality of their wireless transmission path regardless of the specific system (e.g., Wi-Fi, 4G (LTE)/5G), and communication will inevitably be on a best-effort basis. In other words, it will be necessary to prepare for the possibility of momentary communication disruptions. For practical use, to prevent robots from harming people a fail-safe mechanism is absolutely essential, not just in communications but in the entire system.

Artificial Muscle

Artificial muscle is an attempt to replicate human muscle using machines or synthetic materials. Various methods have been developed, but a version suited for service robots has yet to surface. Common drawbacks include the need for large air-pressure generators and the weak strength produced. Among the methods currently under development, artificial muscles made of conductive polymer materials are attracting interest. For example, Toyoda Gosei and Advanced Softmaterials have developed e-Rubber, a next-generation rubber material that expands and contracts when an electric current is passed through the material. With artificial muscles made of e-Rubber, the rubber itself expands and contracts enabling robotic arms to make soft and smooth movements in contrast with other methods like electric motors. e-Rubber has also been successfully applied as a sensor. In the future, the two companies aim to build actuators and sensors using the same material.

Figure 5

Robot Using e-Rubber

Note: Example of a robot for which e-Rubber is used as a pressure sensor.
Robo-Barista exhibited at RoboDEX 2020. The left image shows a panoramic view of Robo-Barista, and the right is an enlarged view of the tactile hand.

Source: Toyoda Gosei Co., Ltd. “Toyoda Gosei to Exhibit e-Rubber at RoboDEX” Toyoda Gosei Co., Ltd. Website
https://www.toyoda-gosei.com/news/detail/?id=247 (Accessed: January 29, 2020)

Future Prospects

With both a long history and a proven track record, most of the technologies required for industrial robots are already in practical use. In contrast, most of the technologies required for service robots, which work and coexist with humans, are still under development. We foresee these technologies reaching practical use in five to ten years and a more advanced level in twenty years.

With such a wide variety of technologies required for service robots, one primary task will be facilitating the necessary collaboration between companies, universities, and national research institutions. In Japan, broad collaboration, as opposed to the typical self-reliant mindset, will be key to developing practical service robots.

While Japan is no longer a leader in the virtual realm, robots require careful tuning to work and coexist with humans and offer an area in which Japan can demonstrate its strength. In addition to its growing presence in both the real and virtual realms, China has positioned robotics as a field of national importance and is making rapid advancements. If Japan is slow to act, China could very well overtake Japan in this field. The national government and corporations must support Japanese engineers and researchers who must in turn do their best for Japan to become a leader in robotics.

  • URLをコピー Copy