You might have heard a lot about artificial intelligence (AI), but how about its quieter counterpart – intelligence amplification (IA)? Just like AI, IA uses machine learning technology as well.
The only difference is that IA complements and amplifies human intelligence instead of bypassing them. In short, AI seeks to replace humans, whereas IA seeks to assist humans.
Now that the global ageing population is growing, IA can make their life easier and more dignified. The home robot is one of the examples, and Toyota Research Institute (TRI) has been working on this technology. It has recently unveiled some of its latest home robot prototypes in the form of a virtual open house.
Hanging Butler Robot
For houses with limited floor space, technology might come in handy where a gantry robot hanging from the ceiling seems to be a great idea. Inspired by Japanese homes where space is particularly constrained, TRI’s new hanging butler robot can be a perfect kitchen helper.
It takes care of the kitchen chores for the elderly who are unable or less keen to do it themselves. It can perform mundane tasks such as loading a dishwasher, wiping surfaces, and tidying up a mess. This super space-saving robotic kitchen helper can be tucked out of the way when it is not in use.
On the other hand, TRI is also testing its floor-based robot helper with the same functions as its bat-like partner.
Soft Bubble Gripper
Improving robotic perception and manipulation is part of TRI’s efforts to develop home robots. TRI has introduced an improvised version of a robotic arm with soft bubble grippers that mimic our fingertips.
These air-filled and elastic bubbles use internal cameras and tactile sensors to feel and recognize the object that they are holding. It is similar to our fingers digging for keys deep inside our pockets. They are capable of more stable grasps and precise placement.
Furthermore, they can also sense when something is slipping or pulling away from them. These capabilities enable the robot to gently hand someone a full glass of wine or place it on the table without spilling. These tasks would be rather challenging for rigid grippers without any tactile sensing ability.
Virtual Reality (VR) Teaching and Fleet Learning
Cleaning is a necessity. Take wiping as an example. This simple task poses a huge programming challenge for traditional robots. It is not easy to measure the exact pressure to wipe the surfaces efficiently.
Imagine all the surfaces at your home with different types of materials. TRI’s researchers are taking a slightly different approach. They teach their robots via VR technology by showing them the desired tasks using VR controllers.
One of the examples is to teach the robot how to open the refrigerator door. The trainer will show the robot where to place its gripper, how to hook the handle, and how hard to push. The robot will then pass what it has learned to other robots. This is how fleet learning works.
Challenges of Domestic Robots
Every home is unique with different layouts and objects being arranged in various configurations. Subsequently, this makes home one of the most complex environments for robots to master.
Transferring what robots are doing in the lab to real homes is indeed challenging. The robots need to constantly learn and adapt to all these changing environments.
In the above-mentioned VR teaching mode, the robot will associate what it sees through its camera to the exact physical act. But this association is only limited to that specific task, scene, or objects that might be present in the scene.
Recognizing the limitations, researchers are working on existing technology to extend what the robot has learned to another new environment. At the same time, they need to ensure the robot is still capable of adopting new tasks in the new environment as well.
More Information About: