Unlike traditional robots that rely on pre-programmed responses, Spot’s narration and conversations are said to be powered by ChatGPT’s ability to generate text in response to the information it processes. So, there could be tailored and dynamic responses to questions and prompts from the people it interacts with.
This is said to have turned Spot into a talking robot. So, this talking robot could not only converse but also navigate its surroundings, observe objects and describe them. The aim was said to be creating a robot capable of engaging in natural dialogues with people, while providing contextual information, based on what it saw. The robot dog’s mouth was, also, said to have the ability to mimic speech, just like a puppet.
The robot, also, has visual input it receives from the cameras on its body. To answer questions and describe its surroundings, Spot is said to employ something called Visual Question Answering. This is said to enable the robot to understand images, interpret them and answer questions about them.
So, Spot could navigate its environment, identify objects and offer detailed explanations in response to inquiries. The robot dog was said to have taken on a variety of personas, including a British male, a sarcastic American, a disinterested teen and a Shakespearean time traveller.
But, there was a challenge facing the robot dog, which is the same challenge that often plagues LLMs: hallucinations. Hallucinations could cause LLMs to generate information that is incorrect. And while this may seem entertaining and light-hearted, the dog robot could serve practical purposes.
