Apple’s ELEGNT AI Framework Can Help Non-Humanoid Robots Convey Intent Via Movement

Apple researchers have developed a new artificial intelligence (AI) framework that will allow non-humanoid robots to express their intentions and engage with humans. Dubbed ELEGNT, the framework focuses on non-verbal communication via movements, postures, and gestures. Alongside focusing on task completion, the framework specialises in allowing robots to express intention, attention, and emotions with movements. Researchers claimed that this will make working with robots a more immersive and engaging experience for users. The capabilities were also tested with human participants.

Apple Develops Frameworks to Let Robots Express via Movements

In a post, the Cupertino-based tech giant detailed the new AI framework. This is aimed at non-anthropomorphic robots, or robots that do not look like humans (contain limbs and a central head). While humanoid robots have an easier time engaging users due to their familiar physical appearance, working with non-humanoid robots can often feel cumbersome.

To bridge this gap, the ELEGNT framework trains robots to express intention and emotions using movements, postures, and gestures. These movements do not have any impact on the robot's task fulfilment and are added just to make human interaction immersive and engaging.

The framework encompasses the hardware design process as well as adds software-based training. The Apple researchers outline a set of interaction scenario storyboards that includes behaviour sequences in different functions and social-oriented tasks. “Our findings indicate that expression-driven movements significantly enhance user engagement and perceived robot qualities,” researchers claimed. A paper documenting the process has been published in the online pre-print journal arXiv.

In a demonstration, the researchers showcased the expressive capabilities of non-humanoid robots in a lamp-like prototype robot. In a video shared by the company, the lamp (which resembles Pixar's Luxo Jr. character) could follow hand gestures to direct the light at highlighted spots. The movements were also designed in a way that gave the impression that the robot was trying to understand the command, nod in agreement, and complete it quickly.

Apple researchers also tested the movements and expressive capabilities of the robot with 21 participants. The tests included functional tasks as well as social tasks such as playing music or engaging in conversations. One participant noted in the paper that without the playfulness of the movements, interacting with a lamp robot would feel “annoying rather than welcome and engaging.”