New research published in Applied Physics Reviews highlights how improved tactile functionalities can improve robots” ability to sort and determine items. The study introduces a proprietary robot with new sensing capabilities. The researchers put these capabilities to the test via a variety of garbage-handling tasks. The goal was to see how tactile robotics could determine complex items in real-life scenarios; here’s what you need to know.
Robots Improved Senses
It’s not a secret that robots have been getting smarter. These devices continue to integrate advanced sensors and AI algorithms to provide improved performance and capabilities. Already, there are robots that can determine an item or even a scenario based on visual, audio, and circumstantial inputs.
Source – Applied Physics Reviews – Garbage Handling
Robot sensors have continually improved and today’s options are light years ahead of their predecessors in terms of accuracy, weight, and capabilities. These sensors have made it possible for robots to begin taking over complex jobs that require accurate object detection.
AI Improvements
AI learning algorithms have played a crucial role in updating the capacities of these units. AI systems enable developers to make robots much smarter. They can deduce situations and items based on surrounding evidence and outside factors, improving their identification systems greatly. Notably, most sensing improvements have come in the form of AI integrations over the last few years.
Tactile Robots – Garbage Handling
One area of robotics that has engineers excited is tactile sensing devices. When you think of robots, you probably envision a humanoid device scanning the room with visual, thermal, and other sensors. However, it’s less common for people to contemplate how robots touch and feel items.
Tactile robots have grown in popularity over the years due to their specific skill sets. These devices can fill vital roles within the workforce and accomplish tasks that visual-oriented devices are incapable of completing. Tactile robots combine various touch sensors to determine the weight, shape, density, and other vital data on an object.
Problems with Tactile Robots Today
There are some serious problems that robotic engineers need to overcome when creating tactile robots. For one, it is extremely difficult to program a robot that has grasping recognition. The human hand has billions of sensors that work together, enabling you to determine if the item you picked up was an apple or a tennis ball.
In real-life scenarios where high accuracy is required, tactical robots haven’t been able to keep up with their accuracy requirements. Issues such as confusing items with similar shapes and sizes are common in these units. Also, background interference and distractions can make it more difficult to accurately categorize items for tactile bots.
Lastly, the orientation and grip of an item play a huge role in determining its classification. Here, tactile robots must combine their data with other sensors to gain a full picture of the scenario, item, and how to handle it. In this way, tactile bots can be taught to determine items, regardless of whether they have shapes and sizes different from the example used during the AI’s training.
Study – Garbage Handling
The study, “Enhanced Robotic Tactile Perception with Spatiotemporal Sensing and Logical Reasoning for Robust Object Recognition,” goes into detail about how and why researchers decided that tactile sensing robots needed a boost in performance. The paper begins with engineers explaining that up until recently, tactile robotics were limited to mostly weight, pressure, and manipulation sensing.
Spatiotemporal Tactile Sensing
The researchers realized that humans also use thermal sensing to determine an item accurately. Multiple sensors, in collaboration with advanced AI algorithms, provided the perfect combination for improving accuracy and determination speeds. The new protocol enabled devices to integrate visual data alongside thermal readings, surface roughness, contact pressure, and temperatures.
Layered Sensor Improves Garbage Handling Capabilities
The team decided to create a new type of robotic sensor that leveraged multiple layers to improve results. Their creation introduced a variety of material detection systems that would scan and register relevant data from surface contact. Additionally, the robotic hand-integrated pressure sensors are located at the bottom of the device.
These sensors are complemented by thermal sensors located in the middle of the hand. Like a human hand, the sensor sits behind a porous material, enabling it to get accurate readings of items in seconds. Notably, this sensor is very sensitive, enabling it to detect minute temperature changes incurred from handling items.
AI Integration
All of this sensory information could have been lost in space if the researchers had not chosen to create a custom AI learning model to sort and make sense of the data. Their creation leverages an efficient cascade classification algorithm that enables their device to figure out complex real-world items simply through quick manipulation.
The AI algorithm was created to act like the human brain in that it rules out the obvious faults first before dipping into more plausible solutions. This cascade classifier leverages multilayered long short-term memory neural networks to provide accurate results. Each item gets hierarchically identified and categorized by the AI autonomously, adding to its future capabilities along the way.
Garbage Handling Test
To test their theory and creations, the researchers began by finding a task that was traditionally very difficult for tactile bots. They settled on garbage sorting for multiple reasons. For one, garbage is very diverse. You can find all types of material in different conditions when sorting garbage.
Another reason why garbage sorting is the ideal task is that there are often multiple items that are the same. The only differences are in their condition, contents, or other trivial but confusing robot scenarios. Tasks such as sorting empty cartons, scraps, plastic bottles, napkins, sponges, orange peels, and expired drugs have proven to be impossible for earlier iterations of robots. This latest upgrade could lead to bots sorting recyclables, food scraps, hazardous waste, and more in the coming years.
Results
The test results were promising with the robots able to successfully sort 98.85% of the garbage provided. These results could have massive ramifications across the industry as they demonstrate that the new approach to tactile robotics is effective and available to the public. The report highlights how the unit could successfully recognize diverse items and sort them according to their classifications.
The study revealed that the robot was able to sort these devices, even when it was scanning the item for the first time. As such, researchers are confident that their creation could play a vital role in garbage handling, recycling, and many other vital tasks in the coming years.
Benefits
There are several benefits that can be deduced from this study. For one, the introduction of more sensitive tactile bots opens the door for far more robot integration. It has been a dream of engineers to create robots that could arrive at a manufacturing plant, learn from humans, and then take over completing their tasks.
In the past, this approach wasn’t possible because the robots weren’t able to handle the categorization steps for items. The integration of thermal sensors and added AI boosts robots’ capabilities and places them on par with their human counterparts. Now, robots can tell if the apple they are holding is real, fake, hot, cold, or even hollow.
Companies that Have the Potential to Benefit from this Research
There are many firms that could gain significant market positioning if they integrate this research into their offering. The robotics sector continues to expand and as more competitors enter the arena, innovation rises. Here are a few major contenders that could integrate these findings and improve their bottom lines.
ABB is a leading international robotics manufacturer with a reputation for quality. This Zürich-based firm offers a variety of industry-ready robotic options to the market. The firm’s most popular offering includes a mobile robotics platform named Flexi Tub that can navigate complex work environments alongside human counterparts.
The Yumi bot is the firm’s latest release. This assembly line bot includes a free-moving arm and a host of sensors that make it ideal for use in conducting line activities such as installations and more. The introduction of thermal, tactile sensors and other items from the researcher’s report would greatly improve this bot’s capabilities. As such, ABBY is a stock worth following.
Zebra Technologies is based out of the US. It offers a variety of industrial robots that can handle complex tasks such as transporting items through busy warehouses while avoiding obstacles and other workers. The Fetch Roller 100 and Fetch 100 shelves are both currently in use by academics.
The main advantages of the bots are their flexibility, easy programmability, and the capability to coordinate tasks via a cloud-based protocol. This added convenience makes tracking and enhancing robot performance within your operations easy. Enhancing tactile sensory capabilities would improve their options, opening the door for more use-case scenarios across industries.
Zebra Technology’s market positioning and the fact that its working prototypes are in use around the globe add to the platform’s stock demand. Analysts predict this stock will see growth as the robotics industry expands. For now, ZBRA is a strong “hold.”
Future
The future for tactile robotics is bright. These devices will help to streamline manufacturing, reduce waste, and help you make it through your daily activities. You can expect to see an increase in AI intelligence and autonomous implementations across the market alongside the rise of these devices. Here are just a few use cases for these advanced devices.
Industrial
Industrial robotics is a billion-dollar sector that continues to expand. Manufacturers have long sought to integrate robots where possible to streamline the manufacturing process. Tactical robots could complete what would otherwise be human tasks, such as picking up an item and squeezing it to test its ripeness.
In the future, robots will be tasked with everything from creating, manufacturing, and testing of new products. As such, their ability to mimic human sensor inputs can help these devices make more informed decisions that reflect the nuances of the human experience while reducing overhead for manufacturers.
Medical
There are many use case scenarios for tactile robots across the medical field. These devices could one day help disabled people with missing limbs feel again. Tactile robotics could hold the key to making prosthetics that feel real and much more.
These scientific breakthroughs will go hand in hand with neural network signals to create a seamless interaction between the patient and their artificial limb. In the coming years, patients will be able to control and feel their prosthetics, just like a real body part. This approach will improve the lives of millions of people around the world.
Agricultural
The agricultural sector is another industry that seeks to integrate tactile robotics as soon as possible. These robots could be used to complete complex tasks such as picking fruit. AI systems could help the unit determine the ripeness of the food and categorize it. Additionally, when you consider the growing number of people living with food uncertainty, robotic agriculture makes perfect sense.
Researchers
The research paper was led by Tsinghua University alumni Qian Mao and Rong Zhu. These engineers wanted to see what it would take to make a tactile robot with the capabilities of the human hand. Their research has proven highly successful and now has the attention of some of the biggest players in the robotics industry.
Trash Sorting Robots are just the Beginning
The garbage-handling capabilities of these robots are just the beginning. Now that engineers have figured out a way to overcome the difficulty of robotic object recognition in complex real-life scenes, it opens the door for a new generation in robotics driven by innovation and a desire to make life easier.
Learn about other cool projects here.