MIT has taken household robots a step further with the introduction of a new motion and task planning system known as PIGINet.
The question of why there aren’t more robots in homes is a complex one, as our homes present unique challenges.
YOU MAY ALSO LIKE: Google Rolls Out ‘Working Location’ Feature For Its Calendar Users
This is in comparison to structured environments like warehouses and factories.
MIT CSAIL researchers are addressing this issue with their new system, PIGINet (Plans, Images, Goal, and Initial facts), which focuses on bringing task and motion planning to home robotic systems.
PIGINet utilizes a transformer encoder, a versatile and advanced model designed to process data sequences.
The system takes in information about the task plan, images of the environment, and symbolic encodings of the initial state and desired goal.
By combining these inputs, the encoder generates predictions about the feasibility of the selected task plan. Currently, the system is primarily focused on kitchen-based activities.
It leverages simulated home environments to create plans that involve interactions with various elements like counters, cabinets, the fridge, sinks, and more.
In simpler scenarios, PIGINet was able to reduce planning time by 80%, while in more complex situations, the reduction was around 20-50%.
The researchers emphasize that the potential applications of PIGINet extend beyond households.
They envision refining the system to suggest alternate task plans when infeasible actions are identified.
This would further expedite the generation of feasible task plans without the need for extensive training datasets.
Ultimately, they believe this could revolutionize the training and application of robots in various homes, offering a more efficient and adaptable approach.
As PIGINet tackles the challenges of home environments and streamlines the planning process, it brings us closer to a future where robots can perform a wider range of tasks in our homes.
By overcoming the complexity and variability of household settings, these advancements have the potential to accelerate the adoption of robotics technology and its integration into our everyday lives.
“The practical applications of PIGINet are not confined to households,” PhD student, Zhutian Yang, said.
YOU MAY ALSO LIKE: GM Plant Workers Union Demand Improved Working Conditions
“Our future aim is to further refine PIGINet to suggest alternate task plans after identifying infeasible actions, which will further speed up the generation of feasible task plans without the need of big datasets for training a general-purpose planner from scratch.
“We believe that this could revolutionize the way robots are trained during development and then applied to everyone’s homes,” he added.
