Title



The collaboration with the National Institute of Advanced Industrial Science and Technology (AIST) is significantly considered as the background that MegaChips decided to start development of frizz. Here we MegaChips, members in charge of developing and marketing frizz, invited Dr. Kurata and Dr. Kourogi from AIST and talked about the present and the future of frizz including the standardization of PDR.
 
Contents
 ▼ Collaboration started when one project came to a standstill.───
 ▼ The PDR+, the advanced form of PDR, will evolve into HDR that is capable of tracking all moving actions.───
 ▼ PDR+ x Social Media Information = Potential for new applications───
 ▼ Discussions on PDR standardization will start at World conference in September. Fair performance evaluation becomes possible.───
 ▼ The time will soon come when PDR becomes a standard feature of smartphone.───
 ▼ What is required for MegaChips───

Top of Page


 Collaboration started when one project came to a standstill.────
Dr. Kurata and Dr. Kourogi, thank you very much for taking time out of your busy schedule for this talk today. The development of the product frizz would have been impossible without the collaboration with AIST. The product has finally gained a certain degree of recognition in the Chinese market, and we have started to receive orders from customers. Do you remember what the original motivation for collaborations was?
Kato

Kurata
Well, if I remember correctly, it started when you came to the exhibition venue. I think that was about five years ago.
We were developing software GPS for mainly digital still cameras in around 2010.We were told by customers that it was meaningless in indoors if they couldn’t obtain location information. However, we could not find the way to solve the issue. We contacted a bunch of people with knowledge on indoor positioning, and asked them who they thought was the leading organization in research related to this field. Most of them said “AIST”, that is why I contacted AIST. Around the same time, I saw an article in Nikkei Electronics written by Dr. Kurata, even with his contact information, which was a pleasant surprise. So I contacted him immediately and we went to the trade show venue to see him.
Kato

Kourogi
At that time, we were giving a demonstration with a PC and sensor module.
Yes you were. We saw the demonstration and were surprised to see that trajectory was being recorded accurately. That was the beginning. Then we started to think how to operate the PDR algorithm, which was being studied by AIST, with low power consumption. The plan to develop software GPS ended up being rejected, and only frizz survived.
Matsumoto

Kurata
The timing happened to be right.
We had just started to think it’d be difficult to succeed with only software.
You approached us at just the right time.
I’m really happy that our collaboration over the years bore fruit.。
We would often work together with research institutions such as universities under consultant contracts. But this is the first time we have collaborated with a particular organization for such a long time.
Matsumoto

Top of Page


 The PDR+, the advanced form of PDR, will evolve into HDR that is capable of tracking all moving actions.───
  ※ HDR : Human Dead Reckoning
I think the performance of the PDR with the 9-axis sensor has improved significantly. What is your next plan?
Kato

Kurata
We come up with many things to work on, such as the behavior measurement of animals. Also, we are recently seriously deal with barometric pressure.
The precision of barometric pressure sensors has improved dramatically, allowing for really innovative applications. We can now correctly recognize actions in stair landing as well as between floors.
Kourogi

Kurata
I agree. A navigation system will be especially useful if it can detect vertical transfers.
It will also be applied by fire departments and police for emergency response, so the height has become an important element. The market will expand rapidly, so we are closely monitoring the trends of applications that utilize barometric pressure sensors.
Kato

Kurata
We are also working on what we call PDR+, although we are still in the initial stages. It seeks to measure activities other than walking by applying activity recognition technologies. Maybe we should call it HDR, or Human Dead Reckoning. Anyway, we believe this is the direction we should be taking.
We travel much longer distance in vehicles and trains than we do on foot. I believe we will definitely need to respond to these demands.
Kourogi

Top of Page


 PDR+ x Social Media Information = Potential for new applications───
Let me digress a bit, (laughs) but I think it’s really interesting to connect car navigation systems with social media information such as Twitter. It’s so engrossing you might get addicted to it. I am sure the car navigation and social media will be linked together.
Kurata

Kato
What do you mean by that?
Naturally we only use car navigation system in the car. However, person’s traffic line is continuous before getting into a car, and after getting out of a car.. In this sense, Dead Reckoning using a smartphone is not necessary in all situations. As long as a smartphone is linked to the network properly, it will be linked to the operation information of a train, or to the navigation system of a car, for example. If we ask ourselves what the next important thing is after the development of IoT and M2M, the answer should be the natural behavior of people. This is what we really need to be able to detect.
Kurata

Matsumoto
I see. This is what you mean by linking with social media information.
In addition, “things” cannot move on their own. People move things, or in some cases forklift moves things. All we need to do is to link these things with somethings that move them, such as people and forklift. As long as we measure things that move, we will also be able to measure it that cannot move. If we can build a network needed for that purpose, we’ll be able to make a classification.
Kurata

Kato
That’s true. We can gather train information through the operation information.
Maybe we can rely on the train operation information only in Japan. (Laughs) As for inaccurate information, however, I believe various services will be created by utilizing social media information.
Kurata
Talking

Top of Page


 Discussions on PDR standardization will start at World conference in September. Fair performance evaluation becomes possible.───
What initiatives do you have in mind for the standardization of PDR, that MegaChips is helping out with?
加藤

Kurata
First of all, the 2015 ACM International Joint Conference on Pervasive and Ubiquitous Computing, and the International Symposium on Wearable Computers (UbiComp / ISWC) will be held in Osaka in September. This will be the first international stage where discussions on standardization will be held. At this conference, people will discuss what data, or dataset as we called it, needs to be collected to determine the evaluation indicators as measures for the data, based on which specifications and performance will be measured. In Japan, the grass roots-level committee called the SIG on PDR Benchmarking Standardization, which you are a part of that was established last year will help out with the discussions.
Now we are seeing a set of two kinds of elements, the dataset and the evaluation index. We intend to focus our efforts on the evaluation index. We’ll first set the index and discuss with those who study the dataset. Of course, it is ideal to have only one comprehensive index, but I believe there may be various approaches regarding evaluation index. For example, we can prepare an index for the low power consumption you’re working on. Attaching importance to the absolute accuracy by assuming the computing costs for combination with the processing on cloud is another possible method. There are also cases where reasonable level of accuracy is more than enough. We are going to collect various index and discuss what is required for PDR with companies and universities.
Kourogi

Kato
I think the approach to evaluation and the index vary according to the customer.
Speaking of evaluations, I think to prevent cheating is also an important matter, I mean taking a shortcut by omitting calculations and setting conditions in advance. It is important to evaluate performance accurately, for example, walk, stop suddenly and walk again, walk holding various things in hand. The dataset needs to include those information, I think it will result in fair index.
Kourogi

Kato
I have heard that there are professional evaluators , some companies only allow professional evaluators to touch the evaluation devices. On the other hand, we just easily lend out demo units to customers for the evaluation, I think that way it is hard to survive in the market.
It sounds fair that each manufacturer determines the method of testing for evaluations. However, the performance differs completely depending on how it is evaluated. People like us, who know the algorithms, will have good results as we know how to walk so that clear locus would appear. So it might be more efficient to let people do the measurement who do not have specialized knowledge..
Matsumoto

Kurata
At the meeting on standardization held the other day, which you were also at, someone suggested an idea to develop human resources who can walk according to the categorized patterns every time and conduct evaluation as the fair testers. Anyway, the measurement is not simple because various human axis, hardware axis of sensors and system and environment are complexly intertwined in the evaluation. What is impressive about standardization activities is that there are many companies taking part. It means that the demands do really exist. We would like to meet their needs to the best of our ability. We would like to start by taking care of the easy things first, instead of focusing too much on academic matters.

Top of Page


 The time will soon come when PDR becomes a standard feature of smartphone.───
In recent years there are people who seriously consider equipping PDR in smartphones. For the indoor positioning systems, the infrastructures were required to be developed firstly such as Wi-Fi and beacon. Through trial and error, I guess people are starting to recognize the limitations of advanced development of infrastructure.
Kato

Kurata
DOCOMO and ZENRIN DataCom just have started indoor navigation service using PDR in April, I think they really took a bold step. They started providing the services in around 320 locations, all at once. This is being done on a massive scale, rather covertly behind the scenes. I think the important point is that they were able to start the services at once because this approach is not dependent on infrastructure. Of course, accuracy will be higher with a hybrid system, so the service will shift in the future. By the time the hybrid system is achieved and PDR comes to be recognized, users will find the battery life is poor. This will definitely be the time for frizz.
Yes, I believe so.
Kato

Kurata
Recently, I’ve gotten my hopes up for smart watches as well. I visit factories and distribution warehouses for the measurements. There have been an increasing number of people recently who desire to use smartphones due to social trend of the “Bring Your Own Device”. However, given the spread of large-sized smartphones, some people find it difficult to constantly carry around their smartphones at work. This has led to the demand for smaller devices like watches, but a lot of focus has been given to the issue of power consumption. For this reason, I feel that a high-performance processor with low power consumption will become absolutely necessary.
Do you mean that PDR will also be mounted in small battery-driven devices in the future, and that LSIs with lower power consumption will be required for these devices?
Kato

Kurata
That’s right. Wireless communication consumes power as well, but if PDR is processed locally, it can be used as the information even if the data is transmitted intermittently, but when the raw data is sent, the data itself becomes different if transmission is interrupted. This is another reason why low power consumption and high-end arithmetic processing are incredibly effective.
It’s true we can use navigation function with watch, it is good enough to show arrows like compass. Actually there is an increasing demand from customers who want to PDR in the watches.
Kato

Matsumoto
For those who cannot read the map, when they know vaguely where they are, it might be easier to obtain information which direction to head into and just walk straightforwardly to it until they finally reach their destination. The application that shows users for example, that the destination must on their right hand side, would be good enough, I think.
It would be good to have a watch that shows an arrow to tell users to make a turn.
Kato

Top of Page


 What is required for MegaChips───
  Development of hardware that achieves high-end arithmetic processing and low power consumption, and improvement of user-friendly development environment through software library maintenance
We’re coming to the end of this round-table talk. Dr. Kurata and Dr. Kourogi, could you tell us what you expect from MegaChips in the future?
Kato

Kurata
OK, I’ll give you a rough idea and Dr. Kourogi will explain details. (Laughter) Anyway, in the midst of this ongoing globalization, we would like to develop fully Japan made hardware including MegaChips’ product.
What we expect from MegaChips are devices of low power consumption. I don’t know whether it is a problem with LSI or the hardware as a whole. What is clear to me is that you cannot wear a large measuring instrument, it must be something like the employee ID card that Mr. Matsumoto is now wearing on his chest pocket.
Kourogi

Kurata
That’s right. That’s exactly what we want.
Everyone wears it, so I believe that form is ideal. The demands for employee ID card with PDR function is increasing nowadays.
Kourogi

Kourogi
There are two factors required for achieving it. The first, of course, lower power consumption of the device is a top priority. Earlier I mentioned about wireless intermittent transmission. Using wireless transmission instead of accumulation allows you to get the data on a real-time basis, and makes it possible to save power as well. So we would like you to continue to develop hardware that can effectively achieve power saving as you do with your frizz. That’s what we do want.
We hope to come up with more ideas about how to reduce the power consumption in terms of the entire system, not to mention sensor processing.
Kato

Kourogi
The other factor is software. Naturally we’re willing to help you with this as much as we can. For example, we can create a power-saving algorithm or help you with the implementation. I believe your hardware and our software are the two wheels of a cart.. It’s simply not enough to only develop hardware, the software must also catch up. I believe this is where the value of our collaboration lies.
We aim to provide easy-to-use algorithms, in addition to the hardware. We must assume that we utilize inexpensive and sensor having much noise, not like the superior ones used for airplanes. This means that we must work on corrections for not only the sensor, but by applying other means as well. In the end the correction will be performed as a whole system. Therefore, even if we sacrifice the certain amount of accuracy, we aim to create software that the errors always remain within the allowable range. As for the hardware, we’ll keep pursuing lower power consumption as we stated earlier.
Matsumoto

Kato
Great. Well, thank you very much for today’s discussion. I really appreciate your time today.
Thank you very much. All
handshake

Top of Page

■Profile



Takeshi Kurata
National Institute of Advanced Industrial Science and Technology(AIST)

Ph.D. (Engineering)
 
Group Leader
Service Sensing, Assimilation, and Modeling Research Group Human Informatics Research Institute
National Institute of Advanced Industrial Science and Technology (AIST)
http://www.aist.go.jp/index_en.html

Professor
(Cooperative Graduate School Program)
Service Engineering and Mixed Reality Laboratory
http://staff.aist.go.jp/t.kurata/lab/index.html
Faculty of Engineering, Information and Systems University of Tsukuba

Membership of academic societies
Society for Serviceology
IEEE (The Institute of Electrical and Electronics Engineers, inc.)
IEICE (Institute of Electronics, Information and Communication Engineers)
VRSJ (The Virtual Reality Society of Japan)
IPSJ (Information Processing Society of Japan)

Takeshi Kurata
http://staff.aist.go.jp/t.kurata/

Service Sensing, Assimilation, and Modeling Research Group
https://unit.aist.go.jp/hiri/seam/en/

Human Informatics Research Institute
https://unit.aist.go.jp/hiri/en/





Masakatsu Kourogi
Site Sesing Inc,

Ph.D (Information Science)

2015.5ー
Director CTO, Site Sesing Inc, (the 82nd Certified AIST start-up company)

2001.4 - 2015.5
National Institute of Advanced Industrial Science and Technology (AIST)


■ interlocutor from MegaChips
Tomoshige Kato
(Marketing of frizz)
MegaChips

Mahito Matsumoto
(Manager of frizz development team)
MegaChips