Teaching robots to respond to natural-language commands

Technique that relies on inverse reinforcement learning, or learning by example, improves task completion rate by 14% to 17% in simulations.

If general-purpose household robots ever become a reality, it would be nice to address them in natural language — to say to a robot, for instance, “Take the dirty dishes to the kitchen.”

Natural-language commands, however, introduce a new layer of complexity to the control of robotic systems, since the same sequence of actions can correspond to many different natural-language commands (“Can you clear the dishes from the dining room?”).

In a paper that my colleagues and I presented last week at the annual meeting of the Association for the Advancement of Artificial Intelligence (AAAI), we apply some of what we’ve learned working on natural-language understanding to the problem of natural-language robotic control.

In particular, we consider the case of inverse reinforcement learning (IRL), in which an AI agent learns to perform a specified task by observing human demonstrations. We augment the standard IRL framework, though, by specifying the agent’s goals in natural language, not, explicitly, as unique states.

IRL methodology.png
A diagram of the researchers’ training methodology, which alternates between updating an autonomous agent’s policy — a set of actions (a) to take in various states (s) in order to achieve a goal (G) — and training a discriminator to recognize the reward function implicit in experts’ examples. The discriminator learns from both positive and negative examples. Some negative examples (sampled trajectories) are relabeled (relabeled trajectories) and used to augment the experts' examples, both for updating the policy and for training the discriminator.
Credit: Glynis Condon

In experiments involving a benchmark data set consisting of high-quality 3-D simulations of an indoor environment, we compared our method to four leading approaches to IRL. 

In cases in which the agent is tested in an environment that it saw during training, our method improves its success rate in achieving goals specified in natural language by 14%, relative to the best-performing baseline. In novel test environments — environments unseen during training — our method improves the agent’s success rate by 17%.

In the paper, we also present a method whereby a trained AI agent deployed to an unfamiliar environment can generate its own training examples tailored to that environment. This additional self-supervised learning improves the agent’s success rate by an additional 38%.

Inverse reinforcement learning

Reinforcement learning is a paradigm in which an agent learns through trial and error. More specifically, it has a reward function — a measure of how successful it is at achieving some goal — and it learns a set of behaviors that maximize its reward.

In inverse reinforcement learning, by contrast, the agent is presented with a set of demonstrations — the examples of a human expert or other agent — and it must learn the reward function implicitly maximized by the experts.

Demonstrations are represented as trajectories, which consist of sequences of alternating states (of the environment and the agent’s place in it) and actions. With IRL, as with standard reinforcement learning, the agent’s ultimate aim is to learn a policy, which dictates which actions to take in which states. With IRL, however, the agent must learn the reward function and the policy simultaneously.

A common approach to IRL is to use a generative adversarial network, or GAN. The training data for the agent is a set of true trajectories, modeled by experts, which accomplish the goal to be learned.

But the training setup also includes an adversarial generator, which creates false trajectories, and the IRL discriminator must learn to distinguish the two. That is, it must learn a reward function that assigns a high value to true trajectories and a low value to false ones. Simultaneously, the adversarial generator tries to learn a policy that generates high-reward trajectories.

We vary this setup by combining each trajectory with an additional input: a natural-language specification of the goal. A single trajectory may have multiple natural-language goals, corresponding to multiple states and actions in the sequence: for example, “go down the hall”, “turn left”, “find the first doorway on your right”, and so on.

In this setting, the negative examples generated by the adversarial generator are trajectories with misaligned natural-language goals: the trajectory maps out a right turn, for instance, but the natural-language goal is “turn left”.

We alternate between using training examples to teach the agent the reward function and to update the agent’s policy. The reward function is trained on both trajectories and natural-language goals (NL goals), and its training data includes negative examples from the adversarial generator. For policy updates, the agent receives only the NL goals — and only from positive examples — and must predict the associated trajectories.

In our experiments, this basic model offered little improvement over existing IRL models, requiring several additional features to improve its performance.

Data augmentation

First, using our expert-supplied trajectories, we trained a variational goal generator to predict NL goals on the basis of trajectories. That model includes a variational autoencoder, a neural network that produces a highly compressed vector representation of each NL goal. The compressed representation captures semantic information about the NL goal, but it loses information about the goal’s phrasing. Re-expanding such a representation produces a new NL goal that is phrased differently from the original but preserves the semantic content.

We use these trajectories with rephrased NL goals as new positive training examples. This augments our supply of expert training examples, which tend to be scarce, increasing robustness through lexical variance.

When a negative example from the adversarial generator — whose NL goal is inaccurate — passes through the label prediction model, the result is a reconstructed trajectory with a correct NL goal. These relabeled trajectories are added to our supply of positive examples as well.

We use our added positive examples to both train the reward function and update the agent’s policy. Not only does this improve the accuracy of the reward function, but it also increases the agent’s ability to generalize to new settings, since it has more varied encounters with the environment to learn from than it would otherwise.

Finally, we explore an additional method for bootstrapping an agent that is asked to perform tasks in an unfamiliar environment. First, the agent learns a new, goal-agnostic policy from existing training data. This policy encodes general principles, such as not trying to move through closed doors. 

Then we use that general policy to generate sample trajectories in the new environment; these pass through the variational goal generator, which assigns them NL goals. We treat these newly labeled trajectories as expert examples in the new setting, and we use them to update the reward function. 

This added layer of training is what increased our agents’ success rates by 36% when they were deployed to new environments. We think this kind of adaptability will be crucial to household robots of the future, which will need to adjust to new environments — when a family moves or goes on vacation, for instance — without being retrained from scratch.

Research areas

Related content

GB, Cambridge
Our team builds generative AI solutions that will produce some of the future’s most influential voices in media and art. We develop cutting-edge technologies with Amazon Studios, the provider of original content for Prime Video, with Amazon Game Studios and Alexa, the ground-breaking service that powers the audio for Echo. Do you want to be part of the team developing the future technology that impacts the customer experience of ground-breaking products? Then come join us and make history. We are looking for a passionate, talented, and inventive Applied Scientist with a background in Machine Learning to help build industry-leading Speech, Language, Audio and Video technology. As an Applied Scientist at Amazon you will work with talented peers to develop novel algorithms and generative AI models to drive the state of the art in audio (and vocal arts) generation. Position Responsibilities: * Participate in the design, development, evaluation, deployment and updating of data-driven models for digital vocal arts applications. * Participate in research activities including the application and evaluation and digital vocal and video arts techniques for novel applications. * Research and implement novel ML and statistical approaches to add value to the business. * Mentor junior engineers and scientists. We are open to hiring candidates to work out of one of the following locations: Cambridge, GBR | London, GBR
GB, Cambridge
Our team undertakes research together with multiple organizations to advance the state-of-the-art in speech technologies. We not only work on giving Alexa, the ground-breaking service that powers Echo, her voice, but we also develop cutting-edge technologies with Amazon Studios, the provider of original content for Prime Video. Do you want to be part of the team developing the latest technology that impacts the customer experience of ground-breaking products? Then come join us and make history. We are looking for a passionate, talented, and inventive Senior Applied Scientist with a background in Machine Learning to help build industry-leading Speech, Language and Video technology. As a Senior Applied Scientist at Amazon you will work with talented peers to develop novel algorithms and modelling techniques to drive the state of the art in speech and vocal arts synthesis. Position Responsibilities: * Participate in the design, development, evaluation, deployment and updating of data-driven models for digital vocal arts applications. * Participate in research activities including the application and evaluation and digital vocal and video arts techniques for novel applications. * Research and implement novel ML and statistical approaches to add value to the business. * Mentor junior engineers and scientists. We are open to hiring candidates to work out of one of the following locations: Cambridge, GBR | London, GBR
CN, 11, Beijing
Are you interested in applying your strong quantitative analysis and big data skills to world-changing problems? Are you interested in driving the development of methods, models and systems for strategy planning, transportation and fulfillment network? Are you interested to cooperate with Amazonians around the world? If so, then this is the job for you. Our team, ATE(Analytics Technology and Engineering) is looking for an Applied Scientist to join our growing Science Team in Bangalore (India)/ Beijing(China). We are responsible for creating core analytics tech capabilities, quantative models, platforms development, and data engineering. We develop scalable analytics applications and research models to optimize operations processes. We standardize and optimize data sources and visualization efforts across geographies, build up, and maintain the online business intelligence services and data mart. You will work with other scientists, professional data engineers, business intelligence engineers, and product managers using rigorous quantitative approaches to ensure high quality data tech products for our customers around the world, including India, Australia, Brazil, Mexico, Singapore and Middle East. Amazon is growing rapidly and because we are driven by faster delivery to customers, a more efficient supply chain network, and lower cost of operations, our main focus is in the development of strategic models and automation tools fed by our massive amounts of available data. You will be responsible for building these models/tools that improve the economics of Amazon’s worldwide fulfillment networks in different countries as Amazon increases the speed and decreases the cost to deliver products to customers. You will work on large-scale vehicle routing and scheduling problems under complex operational and physical constraints. You will also identify and evaluate opportunities to reduce variable costs by improving fulfillment center processes, transportation operations and scheduling, and the execution of operational plans. Finally, you will help create the metrics to quantify improvements to the fulfillment costs (e.g., transportation and labor costs) resulting from the application of these optimization models and tools. Key job responsibilities - Design and develop complex mathematical, simulation and optimization models and apply them to define strategic and tactical needs and drive the appropriate business and technical solutions in the areas of vehicle routing, inventory management, network flow, supply chain optimization, demand planning. - Apply theories of mathematical optimization, including linear programming, combinatorial optimization, integer programming, dynamic programming, network flows and algorithms to design optimal or near optimal solution methodologies to be used by in-house decision support tools and software. - Translating business questions and concerns into specific analytical questions that can be answered with available data using Statistical and Machine Learning methods. - Prototype models by using modeling and programming languages with efficient data querying and modeling infrastructure. - Communicate proposals and results in a clear manner backed by data and coupled with actionable conclusions to drive business decisions. - Collaborate with colleagues from multidisciplinary science, engineering and business backgrounds. - Manage your own process. Prioritize and execute on high impact projects, triage external requests, and ensure to deliver projects in time. We are open to hiring candidates to work out of one of the following locations: Beijing, 11, CHN
GB, Cambridge
Our team builds generative AI solutions that will produce some of the future’s most influential voices in media and art. We develop cutting-edge technologies with Amazon Studios, the provider of original content for Prime Video, with Amazon Game Studios and Alexa, the ground-breaking service that powers the audio for Echo. Do you want to be part of the team developing the future technology that impacts the customer experience of ground-breaking products? Then come join us and make history. We are looking for a passionate, talented, and inventive Applied Scientist with a background in Machine Learning to help build industry-leading Speech, Language, Audio and Video technology. As an Applied Scientist at Amazon you will work with talented peers to develop novel algorithms and generative AI models to drive the state of the art in audio (and vocal arts) generation. Position Responsibilities: * Participate in the design, development, evaluation, deployment and updating of data-driven models for digital vocal arts applications. * Participate in research activities including the application and evaluation and digital vocal and video arts techniques for novel applications. * Research and implement novel ML and statistical approaches to add value to the business. * Mentor junior engineers and scientists. We are open to hiring candidates to work out of one of the following locations: Cambridge, GBR | London, GBR
US, WA, Bellevue
Do you enjoy solving challenging problems and driving innovations in research? Are you seeking for an environment with a group of motivated and talented scientists like yourself? Do you want to create scalable optimization models and apply machine learning techniques to guide real-world decisions? Do you want to play a key role in the future of Amazon transportation and operations? North America Sort Centers (NASC) are experiencing growth and looking for a skilled, highly motivated Research Scientist in partnership with the Modeling and Optimization (MOP) team. The Sort Center network is the critical Middle-Mile solution in the Amazon Transportation Services (ATS) group, linking Fulfillment Centers to the Last Mile. The experience of our customers is dependent on our ability to efficiently execute volume flow through the middle-mile network. Key job responsibilities A Research Scientist - provides analytical decision support to Amazon planning teams via applying advanced mathematical and statistical techniques. - collaborates effectively with Amazon internal business customers, and is their trusted partner - is proactive and independent in discovering and resolving business pain-points within a given scope - is able to identify a suitable level of sophistication in resolving the different business needs - is confident in leveraging existing solutions to new problems where appropriate and is independent in designing and implementing new solutions where needed - is aware of the limitations of their proposed solutions and is proactive in communicating them to the business, and advances the application of sciences towards Amazon business problems by bringing new methods, ideas, and practices to the team and scientific community. A day in the life - Your will be developing model-based optimization, simulation, and/or predictive tools to identify and evaluate opportunities to improve customer experience, network speed, cost, and efficiency of capital investment. - You will quantify the improvements resulting from the application of these tools and you will evaluate the trade-offs between potentially competing objectives. - You will develop good communication skills and ability to speak at a level appropriate for the audience, will collaborate effectively with fellow scientists, software development engineers, and product managers, and will deliver business value in a close partnership with many stakeholders from operations, finance, IT, and business leadership. About the team - At the Modeling and Optimization (MOP) team, we use mathematical optimization, algorithm design, statistics, and machine learning to improve decision-making capabilities across WW Operations from first mile to last mile. - We focus on transportation topology, labor and resource planning for fulfillment facilities, routing science, visualization research, data science and development, and process optimization. - We create models to simulate, optimize, and control the fulfillment network with the objective of reducing cost while improving speed and reliability. - We support multiple business lanes, therefore maintain a comprehensive and objective view, coordinating solutions across organizational lines where possible. We are open to hiring candidates to work out of one of the following locations: Bellevue, WA, USA
US, MA, Cambridge
The Artificial General Intelligence (AGI) team is looking for a highly skilled and experienced Senior Applied Scientist, to lead the development and implementation of cutting-edge algorithms and models for supervised fine-tuning and reinforcement learning through human feedback; with a focus across text, image, and video modalities. As a Senior Applied Scientist, you will play a critical role in driving the development of Generative AI (GenAI) technologies that can handle Amazon-scale use cases and have a significant impact on our customers' experiences. Key job responsibilities - Collaborate with cross-functional teams of engineers, product managers, and scientists to identify and solve complex problems in GenAI - Design and execute experiments to evaluate the performance of different algorithms and models, and iterate quickly to improve results - Think big about the arc of development of GenAI over a multi-year horizon, and identify new opportunities to apply these technologies to solve real-world problems - Communicate results and insights to both technical and non-technical audiences, including through presentations and written reports - Mentor and guide junior scientists and engineers, and contribute to the overall growth and development of the team We are open to hiring candidates to work out of one of the following locations: Bellevue, WA, USA | Cambridge, MA, USA | New York, NY, USA | Sunnyvale, CA, USA
US, MA, North Reading
We are looking for experienced scientists and engineers to explore new ideas, invent new approaches, and develop new solutions in the areas of Controls, Dynamic modeling and System identification. Are you inspired by invention? Is problem solving through teamwork in your DNA? Do you like the idea of seeing how your work impacts the bigger picture? Answer yes to any of these and you’ll fit right in here at Amazon Robotics. We are a smart team of doers that work passionately to apply cutting edge advances in robotics and software to solve real-world challenges that will transform our customers’ experiences in ways we can’t even imagine yet. We invent new improvements every day. We are Amazon Robotics and we will give you the tools and support you need to invent with us in ways that are rewarding, fulfilling and fun. Key job responsibilities Applied Scientists take on big unanswered questions and guide development team to state-of-the-art solutions. We want to hear from you if you have deep industry experience in the Mechatronics domain and : * the ability to think big and conceive of new ideas and novel solutions; * the insight to correctly identify those worth exploring; * the hands-on skills to quickly develop proofs-of-concept; * the rigor to conduct careful experimental evaluations; * the discipline to fast-fail when data refutes theory; * and the fortitude to continue exploring until your solution is found We are open to hiring candidates to work out of one of the following locations: North Reading, MA, USA | Westborough, MA, USA
IL, Tel Aviv
Come build the future of entertainment with us. Are you interested in helping shape the future of movies and television? Do you want to help define the next generation of how and what Amazon customers are watching? Prime Video is a premium streaming service that offers customers a vast collection of TV shows and movies - all with the ease of finding what they love to watch in one place. We offer customers thousands of popular movies and TV shows including Amazon Originals and exclusive licensed content to exciting live sports events. We also offer our members the opportunity to subscribe to add-on channels which they can cancel at anytime and to rent or buy new release movies and TV box sets on the Prime Video Store. Prime Video is a fast-paced, growth business - available in over 240 countries and territories worldwide. The team works in a dynamic environment where innovating on behalf of our customers is at the heart of everything we do. If this sounds exciting to you, please read on. We are looking for an Applied Scientist to embark on our journey to build a Prime Video Sports tech team in Israel from ground up. Our team will focus on developing products to allow for personalizing the customers’ experience and providing them real-time insights and revolutionary experiences using Computer Vision (CV) and Machine Learning (ML). You will get a chance to work on greenfield, cutting-edge and large-scale engineering and science projects, and a rare opportunity to be one of the founders of the Israel Prime Video Sports tech team in Israel. Key job responsibilities We are looking for an Applied Scientist with domain expertise in Computer Vision or Recommendation Systems to lead development of new algorithms and E2E solutions. You will be part of a team of applied scientists and software development engineers responsible for research, design, development and deployment of algorithms into production pipelines. As a technologist, you will also drive publications of original work in top-tier conferences in Computer Vision and Machine Learning. You will be expected to deal with ambiguity! We're looking for someone with outstanding analytical abilities and someone comfortable working with cross-functional teams and systems. You must be a self-starter and be able to learn on the go. About the team In September 2018 Prime Video launched its first full-scale live streaming experience to world-wide Prime customers with NFL Thursday Night Football. That was just the start. Now Amazon has exclusive broadcasting rights to major leagues like NFL Thursday Night Football, Tennis major like Roland-Garros and English Premium League to list few and are broadcasting live events across 30+ sports world-wide. Prime Video is expanding not just the breadth of live content that it offers, but the depth of the experience. This is a transformative opportunity, the chance to be at the vanguard of a program that will revolutionize Prime Video, and the live streaming experience of customers everywhere. We are open to hiring candidates to work out of one of the following locations: Tel Aviv, ISR
MX, DIF, Mexico City
Are you a data enthusiast? Does the world’s most complex logistic systems inspire your curiosity? Is your passion to navigate through hundreds of systems, processes, and data sources to solve the puzzles and identify the next big opportunity? Are you a creative big thinker who is passionate about using data and optimization tools to direct decision making and solve complex and large-scale challenges? Do you feel like your skills uniquely qualify you to bridge communication between teams with competing priorities? If so, then this position is for you! We are looking for a motivated individual with strong analytic and communication skills to join the effort in evolving the network we have today into the network we need tomorrow. Amazon’s extensive logistics system is comprised of thousands of fixed infrastructure nodes, with millions of possible connections between them. Billions of packages flow through this network on a yearly basis, making the impact of optimal improvements unparalleled. This magnificent challenge is a terrific opportunity to analyze Amazon’s data and generate actionable recommendations using optimization and simulation. Come build with us! In this role, your main focus will be to perform analysis, synthesize information, identify business opportunities, provide project direction, and communicate business and technical requirements within the team and across stakeholder groups. You consider the needs of day-to-day operations and insist on the standards required to build the network of tomorrow. You will assist in defining trade-offs and quantifying opportunities for a variety of projects. You will learn current processes, build metrics, educate diverse stakeholder groups, assist science groups in initial solution design, and audit all model implementation. A successful candidate in this position will have a background in communicating across significant differences, prioritizing competing requests, and quantifying decisions made. The ideal candidate will have a strong ability to model real world data with high complexity and delivery high quality analysis, data products and optimizations models for strategic decision. They are excited to be part of, and learn from, a large science community and are ready to dig into the details to find insights that direct decisions. The successful candidate will have good communication skills and an ability to speak at a level appropriate for the audience, will collaborate effectively with scientists, product managers and business stakeholders. Key job responsibilities Statistical Models (ML, regression, forecasting, ) Optimization models, AB and hypothesis testing, Bayesian models. Communication skills with both tech and non tech stakeholders. Writting skills, capable to create documents for different types of readers (business, science, tech) to communicate results on analysis, testing. A day in the life We are open to hiring candidates to work out of one of the following locations: Mexico City, DIF, MEX
US, WA, Seattle
We are seeking a Senior Applied Scientist to join our AI Security team and use AI to develop foundational services that make security mechanisms more effective and efficient. As a Senior Applied Scientist, you will be responsible for researching, modeling, designing, and implementing state-of-the-art AI-based security solutions at Amazon scale. You will collaborate with applied scientists, security engineers, software engineers, as well as internal stakeholders and partners to develop innovative technologies to solve some of our hardest security problems, and build paved path solutions that support builder teams across Amazon throughout their software development journey, enabling Amazon businesses to accelerate the pace of innovation to delight our customers. Key job responsibilities • Research and develop accurate and scalable methods to solve foundational security problems. • Lead and partner with applied scientists and engineers to drive modeling and technical design for complex problems. • Build security tooling and paved path solutions that support builder teams throughout their software development journey. About the team ABOUT AmSec: Diverse Experiences: Amazon Security values diverse experiences. Even if you do not meet all of the qualifications and skills listed in the job description, we encourage candidates to apply. If your career is just starting, hasn’t followed a traditional path, or includes alternative experiences, don’t let it stop you from applying. Why Amazon Security: At Amazon, security is central to maintaining customer trust and delivering delightful customer experiences. Our organization is responsible for creating and maintaining a high bar for security across all of Amazon’s products and services. We offer talented security professionals the chance to accelerate their careers with opportunities to build experience in a wide variety of areas including cloud, devices, retail, entertainment, healthcare, operations, and physical stores. Inclusive Team Culture: In Amazon Security, it’s in our nature to learn and be curious. Ongoing DEI events and learning experiences inspire us to continue learning and to embrace our uniqueness. Addressing the toughest security challenges requires that we seek out and celebrate a diversity of ideas, perspectives, and voices. Training & Career Growth: We’re continuously raising our performance bar as we strive to become Earth’s Best Employer. That’s why you’ll find endless knowledge-sharing, training, and other career-advancing resources here to help you develop into a better-rounded professional. Work/Life Balance: We value work-life harmony. Achieving success at work should never come at the expense of sacrifices at home, which is why flexible work hours and arrangements are part of our culture. When we feel supported in the workplace and at home, there’s nothing we can’t achieve. We are open to hiring candidates to work out of one of the following locations: Austin, TX, USA | San Francisco, CA, USA | Seattle, WA, USA