By Jieming Li (Benjamin) and Jack HsuThe interaction between robots and human is well known to be one of the essential factors to the application of robots in the future. It has great potential in changing both society and the economy in a positive direction since it could build a bridge between machines and humans. One of the critical aspects of these robots is their ability to interact with the environment by making their own decisions. These critical judgments can be made from the robots observing and sensing the environment, which allows them to build social interactions with human workers. But one primary concern is how we can implement these computational algorithms in a manner that it will be safe for workers to work around robots. Due to the risks of exposing human workers to malfunctioned machines, nowadays, robots are mostly separated from humans to ensure safety, and this leads to a decrease in both their flexibility and productivity. So attention has been concentrated at robot-human interaction for production and manufacturing purposes. To get more insights on this topic, we have interviewed Professor Masayoshi Tomizuka — who researches this field at the College of Engineering at UC Berkeley— about the robot-human interaction research he is directing and his opinions of future applications. In a nutshell, we are interested in the topic of the application of robots in an automated factory and how industrial robots collaborate with human beings safely because there are more and more robots deployed to factories and these machines could easily hurt or kill humans with their immense strength and faster speed. Even though robots have enormous potential to accelerate the efficiency of industrial development that is about to change our lives forever, they can only be accepted if they are designed to keep human workers safe. Therefore, we believe that Professor Tomizuka’s rich experience in robotics research can help to reveal much more about the future of factory automation. In the process of the interview, we asked Professor Tomizuka’s research background in the engineering field. More specifically, we asked him about his experience such as the motivation behind, the difficulties he faced, as well as the decisions he made in his research of human-robot interaction. Based on his answers, Professor Tomizuka believes that there is a great potential in the study of robot automation which motivates him to pursue his research further. And one of the biggest motivations is to work with intelligent students because they can come up with great ideas which sometimes could make a massive leap in the development. Another critical element to understand is the issues behind the traditional human-robot interaction and what parts of the problem he wanted to solve by conducting this Safe and Efficient Robot Collaboration System. And his primary goal is to come up with control strategies that are efficient and can make sure human workers work safely. According to Professor Tomizuka, the Safe and Efficient Robot Collaboration System is the next-generation robotic system that can apply to many industries including package delivery, home-service, manufacturing, etc. The reason why it is safe and efficient is that, through sensors, robots are given the abilities not just to sense the environment but also predict peoples’ movement and therefore avoid accidentally crashing into people or hitting things around them. Professor Tomizuka said that the system requires two motion capture cameras as sensors — cameras that can track object’s movement and send relevant data such as velocity and position to robots — so that the robot knows where people are. However, even with multiple sensors, the robot is not able to capture all the information such as uncertainties. One of the critical aspects of the algorithm implemented is to consider the dynamic uncertainty within the environment for the robot to operate efficiently and safely. To account for the uncertainty, researchers perform a parallel control algorithm — two algorithms are running at the same time — and in this case, the two algorithms are for the long term and the short term. The long-term algorithm ensures the efficiency and safety of the robot’s behavior, while the short-term algorithm accounts for instant decision-making. This algorithm can analyze the best way to keep robot safe and efficient at all time while making decisions in a short amount of time, which is an effective way of controlling robots in a uncertain environment. To our surprise, Prof Tomizuka said the system could already sprint that the robot itself seems like a human. He also explained the reason why it can achieve such natural performance. Behind the scenes, the researchers at his Lab applied the MPC framework, which stands for model predictive control — an excellent way to control machines while considering the movement of the body itself in the next period — to control the robots. Therefore, the robots are capable of moving while predicting where the next step of human workers would be. It is this framework that builds the core of this Safe and Efficient Robot Collaboration System. In addition to that, researchers also used multiple depth cameras — a camera that can capture the distances to each object. This equipment not just gives our intelligent robots the way to sense the environment around them but also makes them capable of finding objects’ precise locations in the real world and pick them up correctly. They also used an algorithm to capture human workers’ body and predict what’s going to happen when robots are close to human workers. In this way, robots can react accordingly just like how we interact with others and make sure everything is safe to humans. MPC control technique as described above is a powerful tool especially with the advance of computational powers from computers; however, even with this powerful tool, Professor Tomizuka still thinks that the biggest challenge is that these robots are not intelligent enough. Professor Tomizuka has said, “Controlling robots is not only a technology problem but also it is a human nature problem.” He is trying to say that to control a robot that will be safe and efficient, we not only need to have proper control methods but also how humans program the robots for decision making. The human takes a significant role, whether bad or good, for automated robots. The broader conversation would be how to improve the safety and efficiency of robots further and whether it is ethical to have them work with a human with a potential of hurting them. Professor Tomizuka said, “Engineers pay attention if anything goes wrong and some accident takes place, and if this happens, they say control is the problem.” He is trying to emphasize that to make everything safe with robots, control would be the most prominentregion. With emerging robotic technologies, the boundaries between humans and machines become less and less noticeable. Robots not just starting to look like humans but also work like humans. This trend leads to increased chances that humans would need to interact with machines and even collaborate with them since robots have significant advantages in many aspects compared to humans. This collaboration is an inevitable trend and would make our world developed much faster than before. However, as mentioned in the beginning, safety and efficiency are the main focuses when we interact with robots. The research of Professor Tomizuka and his lab in making the Safe and Efficient Robot Collaboration System that can apply to any kinds of robots is crucial to achieving this goal. Even though they have made tremendous progress in developing this system, as Professor Tomizuka said, there is still a long way to go, and he is optimistic about the future when robots make our world a better place. The work engineers do shapes the world around us. But given the technical nature of that work, non-engineers may not always realize the impact and reach of engineering research. In E185: The Art of STEM Communication, students learn about and practice written and verbal communication skills that can bring the world of engineering to a broader audience. They spend the semester researching projects within the College of Engineering, interviewing professors and graduate students, and ultimately writing about and presenting that work for a general audience. This piece is one of the outcomes of the E185 course.
The future of factory automation was originally published in Berkeley Master of Engineering on Medium, where people are continuing the conversation by highlighting and responding to this story.