The 101st Installment
Active Sensing and Presentation of RobotseTime
by Masahiko Narita,
Professor, Master Program of Information Systems Architecture
IoT, in which collecting a lot of real-world data to store on the internetwork and analyze it, is widely getting attention. I propose to further develop this model to an active sensing model utilizing service robots. In this column, I will be introducing the active sensing model and its system as well as multiple verification tests performed at sightseeing destinations and events, and exploring the presentation of robots.
1. Active sensing model in service robots
It is a model where service robots are used to directly offer services to users and at the same time, robots speak to the users to collect and utilize information about their preferences through methods such as surveys. The characteristic of this is that stakeholders can use the gained information (Picture 1). The surveying techniques have already been researched in depth in the field of psychology so that the results of this can also be utilized. Surveys conducted by robots means that collection of survey data is automated, and that the survey systems can be deployed in the given real space at the same time. Furthermore, we can also expect positive effects such as interaction with a robot, drawing a visitor in and reducing the likelihood a visitor will stop answering in the middle of the survey. Service robots have conventionally been used only to directly offer services to the users, but it was not entirely cost-effective, and it was also not popularized enough. Using the active sensing model will allow us to utilize the data acquired by the robots, as well as to use the service robots in industry fields such as tourism, shopping districts, and events. We believe this has a potential to improve the ecosystem of service robots.
2. Service robots implementation
The high-level function modules such as a conversation function based on the surveys, and applications such as a stamp rally service to utilize the said modules have been implemented on robots and in the cloud, and then built on RSNP (Robot Service Network Protocol). RSNP is a development framework for non-specialists and a protocol for distributed systems for service robots. The surveys are generated on the cloud, and the answers are always stored on the cloud for them to be analyzed and utilized.
The robots used on this system will mainly converse with the users, which means it will most likely not be necessary for them to move. Hence, we have developed a low-cost signage robot (Picture 2), which is a human-size stand with touch panel displays and tablet PC installed. We have also developed a smartphone robot, where the users can answer on their smartphones without installing an app, instead of answering on the robot. As a result, we were able to offer systems for reasonable prices. For the signage robot, because it does not need to move, we prioritized the expressions and movement of the face and response by voice, expressing the presentation as a robot by having it blink, lip sync, and provide backchannels.
3. Verification testing at events and sightseeing destinations
The systems mentioned in (2) have undergone verification tests at large-scale events held at Tokyo Big Sight such as Tokyo International Industry Exhibition 2015, International Robot Exhibition 2015, Japan Robot Week 2016, International Robot Exhibition 2017, as well as international conferences (IIAI AAI 2016 and IIAI AAI 2017) and sightseeing destinations such as Fukagawa Shiryokan dori Shopping Street. As a result, we were able to verify that we could effectively perform surveys at each venue. Also, by implementing sensors to detect changes in numbers and flow of people in the vicinity of the robot, we could understand the response rate of a survey, behavioral patterns such as the time frame when many visitors come, and even the movement patterns within the venue. Therefore, the characteristic of active sensing by using service robots is to collect data on various behaviors of the users within a venue real-time, posing a possibility to perform analysis that goes beyond conventional marketing.
4. Presentation of robots and Kashira Robot
Robots are approachable to the survey respondents and so rather easy to attract potential users. However, signage robots are just faces depicted on 2D displays. In order to attract more potential users more effectively, we developed “Kashira Robot,” which we created with a 3D presentation. Kashira Robot has facial expressions similar to a signage robot, i.e. it blinks, lip syncs, and nods. As for the 3D modeling, if the face of Kashira Robot is not suitably appealing, it will not attract potential users, and might even have the opposite effect, so it must be constructed in an almost flawless manner. We can also assume the necessity to express various faces and characters. In order to meet these requirements, we came up with a design method using a Japanese joruri puppet as a reference, and then modeled using a 3D printer (Picture 3). We received favorable feedback that the appeal of the implementation surpassed expectations from the stakeholders. When it was exhibited at the International Robot Exhibition in 2017, some foreign guests stared at the robot for a long time, and overall, it did not seem to have given any sense of discomfort to the users. We will perform quantitative research on the effects in the future.
5. Summary
In this column, I introduced the active sensing model used in service robots, their concepts and systems, verification testing at events and sightseeing destinations, as well as their presentation and verification. In the future, we intend to work to utilize active sensing in service robots more effectively.