Google AI Researchers Propose TRASS Approach to Create Truly Intelligent Systems

Rukhsar Dhotekar | May 27, 2020

Google AI Researchers Propose TRASS Approach to Create Truly Intelligent Systems
  • A preprint paper published by Stanford University and Google researchers proposes an AI technique that predicts how goals were achieved, effectively learning to reverse-engineer tasks.

  • They say it enables autonomous agents to learn through self-supervision, which some experts believe is a critical step toward truly intelligent systems.

  • A home robot could leverage it to learn tasks like turning on a computer, turning a knob, or opening a drawer, or chores like setting a dining table, making a bed, and cleaning a room.


Learning general policies for complex tasks often requires dealing with unfamiliar objects and scenes, and many methods rely on forms of supervision like expert demonstrations. But these entail significant tuning; demonstrations, for example, must be completed by experts many times over and recorded by special infrastructure.


That’s unlike the researchers’ proposed approach — time reversal as self-supervision (TRASS) — which predicts “reversed trajectories” to create sources of supervision that lead to a goal or goals. A home robot could leverage it to learn tasks like turning on a computer, turning a knob, or opening a drawer, or chores like setting a dining table, making a bed, and cleaning a room.
 

Most manipulation tasks that one would want to solve require some understanding of objects and how they interact. However, understanding object relationships in a task-specific context is non-trivial,” explain the coauthors.

Consider the task [making a bed]. Starting from a made bed, random perturbations to the bed can crumple the blanket, which when reversed provides supervision on how to flatten and spread the blanket. Similarly, randomly perturbing objects in a clean [or] organized room will distribute the objects around the room. These trajectories reversed will show objects being placed back to their correct positions, strong supervision for room cleaning


Read More: GOOGLE TO FACE ANTITRUST CHARGES FOR MONOPOLIZING ONLINE ADS


TRASS works by collecting data given a set of goal states, applying random forces to disrupt the scene, and carefully recording each of the subsequent states. A TRASS-driven agent explores outwardly using no expert knowledge, collecting a trajectory that when reversed can teach the agent to return to the goal states. In this way, TRASS essentially trains to predict the trajectories in reverse so that the trained model can take the current state as input, providing supervision toward the goal in the form of a guiding trajectory of frames (but not actions).
 

At test time, a TRASS-driven agent’s objective is to reach a state in a scene that satisfies certain specified goal conditions. At every step the trajectory is recomputed to produce a high-level guiding trajectory, and the guiding trajectory decouples high-level planning and low-level control so that it can be used as indirect supervision to produce a policy via model and model-free techniques.


In experiments, the researchers applied TRASS to the problem of configuring physical Tetris-like blocks. With a real-world robot — the Kuka IIWA — and a TRASS vision model trained in simulation and then transferred to the robot, they found that TRASS successfully paired blocks it had seen during training 75% of the time and blocks it hadn’t seen 50% of the time over the course of 20 trials each.
 

TRASS has limitations in that it can’t be applied in cases where object deformations are irreversible, for example (think cracking an egg, mixing two ingredients, or welding two parts together). But the researchers believe it can be extended by using exploration methods driven by state novelty, among other things.
 

“Our method … is able to predict unknown goal states and the trajectory to reach them,” they write. “This method used with visual model predictive control is capable of assembling Tetris-style blocks with a physical robot using only visual inputs, while using no demonstrations or explicit supervision.”

Read More: GOOGLE'S AI-POWERED ROBOT CAN GRASP OCCLUDED OBJECTS, DO NEVER-SEEN-BEFORE TASKS
 

Spotlight

Prepare for modern, cloud-native applications and operations with a scaled-out infrastructure. Gain agility to better meet customer needs now and in the future.


Other News

SOFTWARE

Pyxis One Raises $17M to Democratize Codeless AI in Marketing as a SaaS Offering

Pyxis One | September 16, 2021

Pyxis One, a California-based tech startup that aims to revolutionize marketing and consumer intelligence by providing an all-in-one, codeless AI platform, closed $17M in Series B funding. With total funding of $24 million, the tech startup plans to use this round of capital for investments in product development, and to expand to new markets. Celesta Capital and Premji Invest co-led the round with participation from previous investors Chiratae Ventures, pi Ventures, and Exfinity Venture Partners. “Every enterprise is either already rearchitecting themselves to adopt AI, or are dependent on one or more AI-driven platforms,” said Atul Gupta, Partner at Premji Invest...

Read More

AI APPLICATIONS

Enterprise AI platform Dataiku launches managed service for smaller companies

Dataiku | June 15, 2021

Dataiku is going downstream with a new product today called Dataiku Online. As the name suggests, Dataiku Online is a fully managed version of Dataiku. It lets you take advantage of the data science platform without going through a complicated setup process that involves a system administrator and your own infrastructure. If you’re not familiar with Dataiku, the platform lets you turn raw data into advanced analytics, run some data visualization tasks, create data-backed dashboards and train machine learning models. In particular, Dataiku can be used by data scientists, but also business analysts and less technical people. The company has been mostly focus...

Read More

AI TECH

Tech Mahindra and Subex Partner to Drive Scale Adoption of Blockchain-based Solutions for Telecom Operators Globally

Subex | November 05, 2020

To enable fraud mitigation and drive operational efficiencies for communication service providers by reducing compliance complexities and faster time-to-market 5th November 2020, BENGALURU, INDIA – Tech Mahindra, a leading provider of digital transformation, consulting, and business re-engineering services and solutions, and Subex, an industry leader in providing services based on Digital Trust, have announced strategic partnership to roll-out blockchain based solutions for telecom operators globally. These solutions will enable fraud mitigation and drive operational efficiencies for communication service providers (CSP) by reducing compliance complexities and faster tim...

Read More

Google Cloud and STS to Automate U.S. Navy Maintenance Inspections Using AI and ML Technology

Prnewswire | August 28, 2020

Google Cloud and Simple Technology Solutions (STS)—a Google Cloud partner and small business specializing in multi-cloud solutions for the federal government—today announced they are working with the U.S. Navy to modernize the maintenance and repairs inspection process for Navy vessels and facilities. STS will use Google Cloud artificial intelligence (AI) and machine learning (ML) technologies on inspection drone-captured images to detect, prioritize, and predict its maintenance needs. The work was awarded to STS as a Phase I Small Business Innovation Research project due to the technology innovation and potential for commercialization....

Read More

Spotlight

Prepare for modern, cloud-native applications and operations with a scaled-out infrastructure. Gain agility to better meet customer needs now and in the future.

Resources

Events