“Hello World” – Marvin, the robot

The ongoing innovation of AI robots has become quite a fussy topic, related to the exciting expectations on how we envisage something sensational to become reality within our lifetime.

A never ending story- how shall humankind progress towards  future developments and acceptance of AI ? Are future robots going to be able to experience emotions- love, fear, gratefulness or anger ?

Our individual perception towards AI shall create a new reality for our families, acquaintances and overall for future society.

We should discuss this sensitive subject with much respect and hope, with the idea that each of our actions will be a valuable brick in the wall of integrating robots into the society of tomorrow.

Could we find the right questions concerning AI ?

Would they be co-habituating with us, controlling and supporting the energy efficienct use of our bodies ? While we allow them to focus onto the so called ” blue collar “ jobs, we may be able to put our minds into the prosperity of new developments.

Finally in such a reality, would our robot Marvin eventually make a sandwich for us as we ask him, with a smile on his face calling us parents ?

Or in another reality, we make a “battery sandwich “ for him while he uses our names as we had put into his data storage ,to send us on another errand for him?

It has been said that students should always surpass their teachers. Marvin the robot is our creation, our student as we guide it to get accommodated into this world. But would it suit us, or the people around us, for it to surpass its creators in a perspective of not needing us anymore ?

The topic of AI has been attracting much attention recently, having a long history of ups and downs. Looking back behind our shoulders, we have seen a lot of AI “winters” and “summers”.

There were times of abyss and apexes of interest and funding in AI research and its practical applications. Currently, it is the time of culmination and a dramatical rise of pragmatic interest and funding into this area.

This massive fueling of money and hopes is extremely impressive, related to immediate practical solutions, such as robots for tasks that are dangerous for people; highly intelligent and autonomous medical and surgery robots, AI processing enormous amounts of data and machine learning, and other groundbreaking solutions and breakthroughs. Physical appearance and intelligent levels of future robots?

Do you think it will ever be possible to create a robot, which is our complete copy shadowing nature’s most divine creation – human beings ?

The inception of Marvin the robot

Inspired by the numerous possible applications of modern-day AI, we decided to put the idea into practice. All the people from the Ai team in Soft2Run tried their best to contribute to the project.

First, we bought the hardware: the raspberry Pi “portable computer “ and its software, with the camera and additional equipment.

The team was very enthusiastic and helpful, and our management sponsored the whole project and observed its developments with great interest.

The project was split into 2 parts: 1st – offline solutions, including all operations on the Pi, 2nd – online solutions covering detection and recognition on the server, with the response on the Pi.

The next step was to develop the senses of Marvin, namely the “ brain & eyes “. Many options and combinations for object detection and recognition were tested and classified.

As a final stage- combining all the above into a single system, we finished the “ brain “ of Marvin and went to the next challenge – helping Marvin learn to speak.

The greatest obstacle we had to overcome was text to speech recognition in Bulgarian language, which was not developed yet. Therefore, we used the voice of one our colleagues, made trial records and finally Marvin was able to recognize and greet us.

In consideration to offline solutions, an obstacle was the somewhat limited space and functionality of Pi. Still the whole solution requires some optimizations, its means of communication between Pi and server, which the team are continuing to work on.

As it is well known, so far Neural Networks are still somewhat of a black boxe for both scientists and practitioners. Nowadays, convolutional Neural Networks (CNNs) are the current state-of-the-art method for modeling the human brain and its functions.

We believe we have made some modest contributions to this complex area, but this will be the topic of the next article.

How far have we gone and what lies ahead ?

(Keeping up with the hype and moving ahead)

Marvin the robot is an exciting project from numerous perspectives. Firstly, we formed a new but extremely professional team with numerous ideas ,which were interpreted as testing playground for different innovative approaches.

Secondly, Marvin is a good example of putting theory into real-life practice. The first truly representative and entertaining benefit- the robot is able to “see” and greet colleagues in the office, with the future prospect of making small talk and autonomous training and why not to show some “mood” and perform small jobs like an office assistant –  send mails, order food etc.

In a wider perspective, we think that AI has achieved very much so far. AI is not just a system of magic spells but rather a real, applicable tool solving practical problems both for our company and our products and for the world around us.

There are many aspects of life to which AI can contribute significantly. For example: processing huge multidimensional Big Data for the industrial and medical world.

And generally, tackling a lot of problems, which are thought impossible to be solved now – making them quite manageable in the nearest future. To mention just a few, this could cover  “smart devices” like smartphones, smart TV, smart homes and smart cities, and even “smarter” society, which has already been drafted within the projects of Elon Musk and his companies.

Another practical perspective of utilizing AI and Marvin in particular, is the bigger idea of using the unused part of information in our daily world.

In a more futuristic way we are trying to convert “Dark Data” into “ Light” one, it means to exploit huge potentially available data, which is still unused for solving problems in our day-to-day life.

As a final line of thought, these are tempting areas ranging from medicine through entertainment and endless business applications, to develop real systems and why not “android robots” to make our world a much better, safer and ultimately – a more humane place to live.

By: Lana Zhorzholiani