Google’s PaLM-SayCan will make Robots more Friendly to Humans


RobotsGoogle’s PaLM-SayCan: Adds AI-Language Skills to Alphabet’s helper robots

Google has developed a new, large-scale learning model that improves the robot’s overall performance and ability to execute more complex and abstract tasks as well as handle complex requests from people called ‘PaLM-SayCan’ That allows Alphabet-developed robots to better understand the user’s commands and respond correctly. PaLM can help the robotic system process more complex, open-ended prompts and respond to them in ways that are reasonable and sensible

PaLM-SayCan is the first implementation that uses a large-scale AI language model to plan for a real robot. It makes it possible for people to communicate with helper robots via text or speech and improves the robot’s overall performance. PaLM, Google’s LLM, will train its domestic robots in the subtleties of human language so that they grasp the nuances and think for themselves including in complex situations. It is a novel approach that uses advanced language model knowledge to allow a physical agent to follow high-level instructions for physically-based tasks. PaLM-enabled helper robot performing a series of complex tasks using a chain of thought prompting and the step-by-step solution needed to carry out the requests.


Google’s PaLM-SayCan:

Google’s PaLM-SayCan robots can take commands that are safe for a robot to perform and highly interpretable. The robot has a tubular white body with a grasping claw at the end of its arms and the cameras in place of eyes render it a human-like appearance but mostly its functions are robotic with context interpreting capability. It aims to ground the language model in tasks that are practical in a particular real-world context, such as driving or building a car. The new learning model enables the robot to understand the way we communicate, facilitating more natural interaction.

Google says that by integrating PaLM-SayCan into its robots, the bots were able to plan correct responses to 101 user–instructions 84 percent of the time and successfully execute them 74 percent of the time. It enables the robot to understand the way researchers communicate. Google Research and Everyday Robots are working together to combine the best of language models with robot learning. Most robots only respond to short and simple instructions, like “bring me a bottle of water”. But LLMs like GPT-3 and Google’s MuM can better parse the intent behind more oblique commands.

PaLM can help the robotic system process more complex, open-ended prompts and respond to them in reasonable and sensible ways. Google combined PaLM with robotic motor skills, the output is a combination of the best language and robotic skills. When the system was integrated with PaLM, they saw a 14% improvement in the planning success rate. Google Researchers also saw a 13% increase in the execution success rate or successfully carrying out a task.