The Future of Robotics: A Look at Toyota Research Institute's Breakthroughs

As we look to the future of robotics, it's clear that our current understanding is still evolving and often inaccurate. In Silicon Valley's Los Altos, California, researchers at the Toyota Research Institute (TRI) are working tirelessly to create robots that can truly make a positive impact on society.

The TRI headquarters serves as a hub for innovation, where experts like Max Bajracharya lead the TRI Robotics team in pushing the boundaries of robotics technology. "A lot of what we think we know about the future of robotics is either still very much TBD or just plain wrong," says Bajracharya. "We're here to sort it all out and explore new possibilities."

One area of focus for TRI is the development of robots that can interact with their environment in a more human-like way. To achieve this, they've been working on soft bubble gripper technology, which allows robots to pick up objects without damaging them or bumping into people. This is made possible by adding softness around the robot's edges and incorporating sensing capabilities.

The idea of replicating human senses in robots is a complex one. As we think about how many sensors our bodies possess and the vast amount of data that needs to be processed, it becomes clear that creating intelligent robots requires significant advancements. "What if you put some softness around the whole robot?" wonders Bajracharya. "Think of our skin as being a little bit soft. What if we actually add sensing to that, replicating our body?"

This is where TRI's research comes in – understanding how to replicate human-like sensory experiences in robots. They've found ways for their robots to handle optical trickery such as clear plastic packages on the grocery shelf and shiny objects that reflect confusing images. "We're looking at the insides of some of the parts," says Bajracharya, referring to the intricate details within components like Wally (likely a nod to the popular toy). The TRI team has developed ways to analyze these internal mechanisms using data and machine learning.

However, as we strive to create more intelligent robots, it's essential to remember that our ultimate goal is not to replace humans but rather augment their capabilities. Bajracharya emphasizes this point: "Our goal is really not to replace people; what you don't see today is humans and robots really interacting together."

In order for these interactions to happen seamlessly, TRI has discovered that simply having humans remotely operate robots is often more work than it's worth. This high cognitive load can be challenging to manage. Instead, the team recommends sharing control with the robot when possible, allowing humans to provide high-level direction while the robot handles specific tasks.

The TRI Robotics team doesn't aim to develop end products but instead focuses on core research and open-sourcing their findings. As a result, these breakthroughs will have a lasting impact on robotics technology and pave the way for future innovations like Toyota's Woven City project.

This cutting-edge research in Silicon Valley holds immense promise for the future of robotics, with potential applications ranging from improving human-robot interactions to enhancing overall efficiency in various industries. As Bajracharya and his team continue their work, we can look forward to seeing these advancements transform the way we live and work.

"WEBVTTKind: captionsLanguage: ena lot of what you and i think we know about the future of robotics is either still very much tbd or just plain wrong let's take a look at a place in silicon valley where they're sorting it all out for our future this is toyota research institute headquarters we're in silicon valley los altos yep california what happens here that people here are passionate about taking robotics technology and having a real impact on people on society making robots truly useful to people to see that we got exclusive access to toyota research institute where max bajracharya leads the tri robotics team and showed me a few of their new breakthroughs like soft bubble gripper technology to create a robot that isn't all hard edges and hard force but can soft sense things like we do allowing it to pick things up without damaging them or bump into you without damaging you well what if you put some softness around the whole robot you know think of our you know our skin is a little bit soft and what if you actually add sensing to that replicating our body is incredibly complex if you think about how many sensors we have and all of that data getting sent up into some processing unit that's one of the really big challenges or how about the diabolical chaos that is the grocery shelf clear plastic packages that appear not to be there to a robot or shiny things that reflect back a confounding image tri has found ways to let their robots handle such optical trickery the same way we figure it out we don't think twice about understanding transparent objects that we can actually see right through them and that our eyes still can figure out what is the depth on that thing that we just saw through what our systems are actually able to do now um using a lot of data and machine learning is they're actually able to use context to understand the what they're looking at in a much more fundamental way we're looking at the insides of some of the parts a lot like wally i think you know that yep right okay just saying all right what you can't see in this tri robot is something that they call ia sort of turning ai on its head it stands for intelligence amplification amplifying our intelligence instead of the artificial intelligence that most of us assume is what robots will use to displace us intelligence amplification just amplification in general of people our goal is really not to replace people what you don't see today is humans and robots really interacting together and that really limits how much robots can really amplify people's ability now you might think a lot of these hurdles can be solved by having humans just remotely operate robots max and his team have found that that approach is often more work than it's worth it's incredibly high cognitive load um if you think of me on my humans yes so what are you gonna do to make that not be the case to take the cognitive load and make it light and bright for me so i'm like super powerful through a robot you as a human are understanding everything about the context of of the scene and you're telling the robot everything how to move what to do but if the robot has some understanding of what it's looking at then all you need to do is give it sort of high level direction you know pick that pick that thing you know like you know put this over there if you're doing that then more of a command than an actual execution exactly that's what's missing this this robot is capable of doing physically capable of doing many of those tasks yeah it's not smart enough the tri robotics team doesn't develop end products they do core research and they open source a lot of what they learn by the way where you will see a lot of this work in reality is that toyota's woven city where other parts of toyota and their partners will take what's learned here and build it into reality therea lot of what you and i think we know about the future of robotics is either still very much tbd or just plain wrong let's take a look at a place in silicon valley where they're sorting it all out for our future this is toyota research institute headquarters we're in silicon valley los altos yep california what happens here that people here are passionate about taking robotics technology and having a real impact on people on society making robots truly useful to people to see that we got exclusive access to toyota research institute where max bajracharya leads the tri robotics team and showed me a few of their new breakthroughs like soft bubble gripper technology to create a robot that isn't all hard edges and hard force but can soft sense things like we do allowing it to pick things up without damaging them or bump into you without damaging you well what if you put some softness around the whole robot you know think of our you know our skin is a little bit soft and what if you actually add sensing to that replicating our body is incredibly complex if you think about how many sensors we have and all of that data getting sent up into some processing unit that's one of the really big challenges or how about the diabolical chaos that is the grocery shelf clear plastic packages that appear not to be there to a robot or shiny things that reflect back a confounding image tri has found ways to let their robots handle such optical trickery the same way we figure it out we don't think twice about understanding transparent objects that we can actually see right through them and that our eyes still can figure out what is the depth on that thing that we just saw through what our systems are actually able to do now um using a lot of data and machine learning is they're actually able to use context to understand the what they're looking at in a much more fundamental way we're looking at the insides of some of the parts a lot like wally i think you know that yep right okay just saying all right what you can't see in this tri robot is something that they call ia sort of turning ai on its head it stands for intelligence amplification amplifying our intelligence instead of the artificial intelligence that most of us assume is what robots will use to displace us intelligence amplification just amplification in general of people our goal is really not to replace people what you don't see today is humans and robots really interacting together and that really limits how much robots can really amplify people's ability now you might think a lot of these hurdles can be solved by having humans just remotely operate robots max and his team have found that that approach is often more work than it's worth it's incredibly high cognitive load um if you think of me on my humans yes so what are you gonna do to make that not be the case to take the cognitive load and make it light and bright for me so i'm like super powerful through a robot you as a human are understanding everything about the context of of the scene and you're telling the robot everything how to move what to do but if the robot has some understanding of what it's looking at then all you need to do is give it sort of high level direction you know pick that pick that thing you know like you know put this over there if you're doing that then more of a command than an actual execution exactly that's what's missing this this robot is capable of doing physically capable of doing many of those tasks yeah it's not smart enough the tri robotics team doesn't develop end products they do core research and they open source a lot of what they learn by the way where you will see a lot of this work in reality is that toyota's woven city where other parts of toyota and their partners will take what's learned here and build it into reality there\n"