Inspired by the seamless ability of humans to handle objects without visual assistance, a team of engineers from the University of California San Diego has successfully created a groundbreaking technology that allows a robotic hand to rotate objects through touch alone, eliminating the need for sight. This revolutionary development holds immense potential for various applications involving robots operating in the absence of light. The image accompanying this article is provided by Binghao Huang.
On July 25, a group of researchers led by engineers from the University of California San Diego unveiled a remarkable robotic hand capable of rotating objects by relying solely on touch-based sensors rather than visual perception.
In a news release, the researchers state that their work has the potential to contribute to the advancements in creating robots that are capable of manipulating objects in dark environments. The purpose behind the need for such robots is still unclear.
The team presented their study at a conference held in South Korea, showcasing the extensive capabilities of their innovative technique. According to the published news release, the robotic hand they developed is capable of smoothly rotating various objects, including small toys, cans, fruits, and vegetables, without causing any damage.
Remarkably, this incredible functionality is achieved using touch sensors that cost a mere $192 in total. Each sensor, priced at $12, is designed to detect whether an object is being touched or not, providing sufficient information to allow the hand’s software to initiate the rotation.
The software utilized in this mechanism is trained through Reinforcement Learning, a popular machine learning paradigm within the field of artificial intelligence.
What sets this approach apart is the reliance on multiple low-cost, low-resolution touch sensors that generate simple binary signals of touch or no touch, enabling the hand to execute in-hand rotation effectively, as highlighted in the news release.
The researchers emphasize the advantages of using these cost-effective sensors, extending beyond their affordability. They clarify that detailed information about an object’s texture is unnecessary for this task, further stating that “simple binary signals of whether the sensors have touched the object or not” are sufficient and easier to simulate and implement in real-life scenarios.
Xiaolong Wang, a professor of electrical and computer engineering at UC San Diego, acknowledges the inherent difficulty machines face in performing in-hand manipulation of objects. This challenge necessitates extensive research and development in the field.
The study begins with a relatable analogy that illustrates the complexity of rotating an object for machines and underscores the significance of conducting research in this area. The paper describes a scenario where someone is washing a pan in the dark due to a sudden power outage. Even with no visual input available, light is used from a mobile phone to proceed with the task. This seemingly straightforward sequence of actions necessitates precise execution of in-hand dexterous manipulation, emphasizing the challenges posed by occlusions and the absence of vision.
Denial of responsibility! VigourTimes is an automatic aggregator of Global media. In each content, the hyperlink to the primary source is specified. All trademarks belong to their rightful owners, and all materials to their authors. For any complaint, please reach us at – [email protected]. We will take necessary action within 24 hours.
