RT-X: An AI Model to Advance Robotic Learning

October 12, 2023


Robotics has traditionally operated within a realm of specialization. Each robot, whether used in an industrial setting or for service-oriented tasks, is specifically designed and programmed for its intended function. However, as we move further into the era of advanced technology, there's an increasing demand for robots that can handle a wider variety of tasks without requiring specialized models for each one1.

The challenge lies in the fact that, historically, a unique model is trained and tailored to each specific robot and its environment or task. Altering any parameter often means retraining the model from scratch, a procedure that's not only resource-intensive but also inefficient. This is the problem that the RT-X initiative aims to address.

The core vision of RT-X is not just the development of a universally adaptable robot. It's about synthesizing knowledge from the entire robotic ecosystem into a unified, scalable model. Through collaborative efforts spanning 33 academic laboratories and incorporating data from 22 distinct robot types, the RT-X project aspires to transform this vision into a practical solution for the future of robotics.

Scaling Up Learning through Open X-Embodiment Repository

With the rapidly advancing pace of technology, robots have found applications in various sectors, from industrial automation to healthcare and domestic chores. However, a significant limitation of most robotics systems lies in their specialization. Each robot, fine-tuned for a specific task, often struggles to adapt to new challenges outside its trained domain. Such narrow specialization inherently restricts the versatility and applicability of robotic solutions.

In light of this, the aspiration to train a robot that can function across lots of tasks, environments, and conditions emerges as a logical next step. The answer lies in pooling collective robotic knowledge. The field of robotics has long been dominated by proprietary systems and isolated datasets. To break this mold and promote a more collaborative approach to robotics research, the Open X-Embodiment Repository was introduced. This open-source repository provides a centralized platform where researchers and developers can access a wealth of data and resources related to diverse robotic platforms.

Central to this repository is the Open X-Embodiment Dataset. This dataset stands out due to its vastness, encompassing over 1 million robot trajectories from 22 different robot types. By consolidating data from 60 existing robot datasets, sourced from 34 leading robotic research labs globally, the Open X-Embodiment Dataset offers an unparalleled breadth of robotic knowledge and actions. To ensure the data remains accessible and user-friendly, it's formatted using the RLDS data format, making it compatible with most deep learning frameworks.

Furthermore, the repository offers pre-trained model checkpoints. These checkpoints, specifically the RT-X model, are designed for researchers to build upon, offering a foundation for further exploration and adaptation.In essence, the Open X-Embodiment Repository isn't just a collection of data. It's a call to the global robotics community, inviting researchers to collaborate, innovate, and push the boundaries of what's possible in the realm of robotic learning.

By moving away from the paradigm of individualized training for each robot type and environment, the RT-X approach offers a promising solution. The ability to train a universal model that learns from multiple embodiments not only reduces the need for repetitive task-specific training but also enhances the adaptability and efficiency of robotic systems as a whole.

Foundations: RT-1 and RT-2 Models

Before diving into the nuances of RT-X, it's essential to understand the foundational models upon which it builds: RT-1 and RT-2.

RT-1: Designed with robotic control in mind, RT-1 operates on a Transformer architecture, a novel approach in machine learning. With a focus on real-world robotic control at scale, this model utilizes visual input and processes it through a pre-trained model like EfficientNet. Alongside visual data, natural language instructions guide the robot's actions, with these instructions transformed into embeddings. The culmination of these processes allows RT-1 to make informed decisions about robot actions in its environment2.

RT-2: Differing from RT-1, RT-2 is a vision-language-action (VLA) model. It merges visual and language data, casting robot actions as natural language tokens. The benefit? By translating actions to text tokens, RT-2 can tap into the vast domain of pre-trained vision-language models, enhancing its versatility. This model's unique approach means it can be fine-tuned for robotic control, making it adept at understanding and executing complex tasks based on visual and linguistic inputs3.

Potential of RT-X 

Historically, each robot was designed with a specific purpose in mind. If a robot was built for industrial manufacturing, its abilities were limited to that environment. However, with RT-X, there's a shift in this paradigm. Instead of training individual models for each specific task or robot type, RT-X promotes the idea of a unified model that's equipped to handle a multitude of tasks across different robotic platforms.

But what makes RT-X stand out? The key is its ability to amalgamate data from diverse robotic platforms. Using the vast Open X-Embodiment Dataset, RT-X is trained across 22 different robot types. This cross-training means the model isn't just limited to the capabilities of one robot type but has insights from a myriad of platforms.

The initial results have been promising. For instance, the RT-1-X variant of the model displayed a significant improvement over traditional, robot-specific models. In evaluations, RT-1-X showed a 50% higher success rate compared to conventional methods. Similarly, the RT-2-X variant demonstrated impressive capabilities, especially in understanding spatial relationships and adapting to varied tasks.

In a nutshell, RT-X isn't just another robotics model. It signifies a step towards a future where robots are not bound by their specific design or initial training but can adapt, learn, and operate across a wide range of scenarios.

Evaluating Emergent Skills with RT-X

In robotics, adaptability remains one of the most sought-after capabilities. While traditional models have been adept at performing tasks they were explicitly trained for, the real challenge lies in their ability to exhibit emergent skills—those not explicitly taught during training but inferred from the data they've been exposed to. RT-X, in its innovative design, promises a step forward in this direction. 

For RT-X, this evaluation was centered around its performance on tasks involving objects and skills that were unfamiliar. For instance, while the Google Robot might have extensive training data from the RT-2 dataset, certain tasks were difficult for it. Yet, when RT-2-X, a variant of RT-X, was put to the test, it showcased a profound ability to adapt and perform. In fact, in comparative evaluations, RT-2-X demonstrated a whopping threefold improvement over RT-2 in emergent skills.

The true strength of RT-X's emergent skill capabilities was further underscored through a series of ablation studies:

Training Diversity: When RT-2-X was trained excluding the Bridge dataset, its performance on hold-out tasks saw a significant dip. This pointed to a vital insight—the diversity of data from various robots played a pivotal role in enhancing emergent skills.

Decoding Design Decisions: Various design elements of RT-2-X were considered including factors like image history, web-based pre-training, and model size. Larger models, such as the 55B variant, displayed a superior ability to transfer knowledge across robotic datasets compared to their smaller counterparts.

Spatial Understanding: One of the standout features of RT-2-X is its advanced spatial understanding. The model could discern subtle linguistic nuances, like the difference between move apple near cloth and move apple on cloth, and modify its actions accordingly. Such an ability underscores the model's potential in understanding and adapting to real-world scenarios.

Results and Findings

The RT-X model, with its unique design considerations and vast training data, has brought forth some compelling results that set new benchmarks in robotics. Here's a breakdown of its primary achievements:

RT-1-X: In comparative evaluations, RT-1-X showcased an impressive 50% improvement over the existing state-of-the-art methods. This significant boost in performance underscores RT-1-X's advanced learning capabilities and adaptability.

RT-2-X: An equally notable achievement was the ability of RT-2-X to generalize across different robotic tasks. When pitted against models that were trained only on data from specific robotic embodiments, RT-2-X exhibited a threefold improvement in generalization capabilities. This makes it a formidable tool in diverse robotic environments, highlighting its versatility.


It is essential to introspect, identify current limitations, and chart the way forward. 

Current Limitations of RT-X

Like any pioneering technology, RT-X isn't devoid of challenges. While it excels in many domains, there are scenarios where its adaptability might be constrained. For instance, it doesn't always consider robots with vastly different sensing and actuation modalities. Also, the model hasn't been fully tested for its generalization to entirely new robots, leaving room for future enhancements.

Future Prospects and the Potential of X-Robot Learning

The horizon looks promising for X-robot learning. RT-X has laid a robust foundation, and the potential of a truly X-embodied robot generalist is within sight. As datasets grow richer and more diverse, and as models become even more sophisticated, the capabilities of such systems are only set to expand. The dream is a future where robots can seamlessly adapt, learn, and perform tasks beyond their initial programming.

While RT-X has broken new ground, the journey of exploration is far from over. To truly realize the dream of universally adaptable robots, there's a pressing need for more research. This encompasses studying how positive transfer can be better optimized, understanding the intricacies of cross-embodiment generalization, and refining models to be even more responsive to diverse robotic tasks.


1. Chebotar, Y. & Yu, T. RT-2: ‘New Model Translates Vision and Language into Action.’ DeepMind https://www.deepmind.com/blog/rt-2-new-model-translates-vision-and-language-into-action (2023).

2. Brohan, A. et al. ‘RT-1: Robotics Transformer for Real-World Control at Scale.’ (2022).

3. Brohan, A. et al. ‘RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Control.’ (2023).

Latest Blogs
This is a decorative image for: A Complete Guide To Customer Acquisition For Startups
October 18, 2022

A Complete Guide To Customer Acquisition For Startups

Any business is enlivened by its customers. Therefore, a strategy to constantly bring in new clients is an ongoing requirement. In this regard, having a proper customer acquisition strategy can be of great importance.

So, if you are just starting your business, or planning to expand it, read on to learn more about this concept.

The problem with customer acquisition

As an organization, when working in a diverse and competitive market like India, you need to have a well-defined customer acquisition strategy to attain success. However, this is where most startups struggle. Now, you may have a great product or service, but if you are not in the right place targeting the right demographic, you are not likely to get the results you want.

To resolve this, typically, companies invest, but if that is not channelized properly, it will be futile.

So, the best way out of this dilemma is to have a clear customer acquisition strategy in place.

How can you create the ideal customer acquisition strategy for your business?

  • Define what your goals are

You need to define your goals so that you can meet the revenue expectations you have for the current fiscal year. You need to find a value for the metrics –

  • MRR – Monthly recurring revenue, which tells you all the income that can be generated from all your income channels.
  • CLV – Customer lifetime value tells you how much a customer is willing to spend on your business during your mutual relationship duration.  
  • CAC – Customer acquisition costs, which tells how much your organization needs to spend to acquire customers constantly.
  • Churn rate – It tells you the rate at which customers stop doing business.

All these metrics tell you how well you will be able to grow your business and revenue.

  • Identify your ideal customers

You need to understand who your current customers are and who your target customers are. Once you are aware of your customer base, you can focus your energies in that direction and get the maximum sale of your products or services. You can also understand what your customers require through various analytics and markers and address them to leverage your products/services towards them.

  • Choose your channels for customer acquisition

How will you acquire customers who will eventually tell at what scale and at what rate you need to expand your business? You could market and sell your products on social media channels like Instagram, Facebook and YouTube, or invest in paid marketing like Google Ads. You need to develop a unique strategy for each of these channels. 

  • Communicate with your customers

If you know exactly what your customers have in mind, then you will be able to develop your customer strategy with a clear perspective in mind. You can do it through surveys or customer opinion forms, email contact forms, blog posts and social media posts. After that, you just need to measure the analytics, clearly understand the insights, and improve your strategy accordingly.

Combining these strategies with your long-term business plan will bring results. However, there will be challenges on the way, where you need to adapt as per the requirements to make the most of it. At the same time, introducing new technologies like AI and ML can also solve such issues easily. To learn more about the use of AI and ML and how they are transforming businesses, keep referring to the blog section of E2E Networks.

Reference Links




This is a decorative image for: Constructing 3D objects through Deep Learning
October 18, 2022

Image-based 3D Object Reconstruction State-of-the-Art and trends in the Deep Learning Era

3D reconstruction is one of the most complex issues of deep learning systems. There have been multiple types of research in this field, and almost everything has been tried on it — computer vision, computer graphics and machine learning, but to no avail. However, that has resulted in CNN or convolutional neural networks foraying into this field, which has yielded some success.

The Main Objective of the 3D Object Reconstruction

Developing this deep learning technology aims to infer the shape of 3D objects from 2D images. So, to conduct the experiment, you need the following:

  • Highly calibrated cameras that take a photograph of the image from various angles.
  • Large training datasets can predict the geometry of the object whose 3D image reconstruction needs to be done. These datasets can be collected from a database of images, or they can be collected and sampled from a video.

By using the apparatus and datasets, you will be able to proceed with the 3D reconstruction from 2D datasets.

State-of-the-art Technology Used by the Datasets for the Reconstruction of 3D Objects

The technology used for this purpose needs to stick to the following parameters:

  • Input

Training with the help of one or multiple RGB images, where the segmentation of the 3D ground truth needs to be done. It could be one image, multiple images or even a video stream.

The testing will also be done on the same parameters, which will also help to create a uniform, cluttered background, or both.

  • Output

The volumetric output will be done in both high and low resolution, and the surface output will be generated through parameterisation, template deformation and point cloud. Moreover, the direct and intermediate outputs will be calculated this way.

  • Network architecture used

The architecture used in training is 3D-VAE-GAN, which has an encoder and a decoder, with TL-Net and conditional GAN. At the same time, the testing architecture is 3D-VAE, which has an encoder and a decoder.

  • Training used

The degree of supervision used in 2D vs 3D supervision, weak supervision along with loss functions have to be included in this system. The training procedure is adversarial training with joint 2D and 3D embeddings. Also, the network architecture is extremely important for the speed and processing quality of the output images.

  • Practical applications and use cases

Volumetric representations and surface representations can do the reconstruction. Powerful computer systems need to be used for reconstruction.

Given below are some of the places where 3D Object Reconstruction Deep Learning Systems are used:

  • 3D reconstruction technology can be used in the Police Department for drawing the faces of criminals whose images have been procured from a crime site where their faces are not completely revealed.
  • It can be used for re-modelling ruins at ancient architectural sites. The rubble or the debris stubs of structures can be used to recreate the entire building structure and get an idea of how it looked in the past.
  • They can be used in plastic surgery where the organs, face, limbs or any other portion of the body has been damaged and needs to be rebuilt.
  • It can be used in airport security, where concealed shapes can be used for guessing whether a person is armed or is carrying explosives or not.
  • It can also help in completing DNA sequences.

So, if you are planning to implement this technology, then you can rent the required infrastructure from E2E Networks and avoid investing in it. And if you plan to learn more about such topics, then keep a tab on the blog section of the website

Reference Links



This is a decorative image for: Comprehensive Guide to Deep Q-Learning for Data Science Enthusiasts
October 18, 2022

A Comprehensive Guide To Deep Q-Learning For Data Science Enthusiasts

For all data science enthusiasts who would love to dig deep, we have composed a write-up about Q-Learning specifically for you all. Deep Q-Learning and Reinforcement learning (RL) are extremely popular these days. These two data science methodologies use Python libraries like TensorFlow 2 and openAI’s Gym environment.

So, read on to know more.

What is Deep Q-Learning?

Deep Q-Learning utilizes the principles of Q-learning, but instead of using the Q-table, it uses the neural network. The algorithm of deep Q-Learning uses the states as input and the optimal Q-value of every action possible as the output. The agent gathers and stores all the previous experiences in the memory of the trained tuple in the following order:

State> Next state> Action> Reward

The neural network training stability increases using a random batch of previous data by using the experience replay. Experience replay also means the previous experiences stocking, and the target network uses it for training and calculation of the Q-network and the predicted Q-Value. This neural network uses openAI Gym, which is provided by taxi-v3 environments.

Now, any understanding of Deep Q-Learning   is incomplete without talking about Reinforcement Learning.

What is Reinforcement Learning?

Reinforcement is a subsection of ML. This part of ML is related to the action in which an environmental agent participates in a reward-based system and uses Reinforcement Learning to maximize the rewards. Reinforcement Learning is a different technique from unsupervised learning or supervised learning because it does not require a supervised input/output pair. The number of corrections is also less, so it is a highly efficient technique.

Now, the understanding of reinforcement learning is incomplete without knowing about Markov Decision Process (MDP). MDP is involved with each state that has been presented in the results of the environment, derived from the state previously there. The information which composes both states is gathered and transferred to the decision process. The task of the chosen agent is to maximize the awards. The MDP optimizes the actions and helps construct the optimal policy.

For developing the MDP, you need to follow the Q-Learning Algorithm, which is an extremely important part of data science and machine learning.

What is Q-Learning Algorithm?

The process of Q-Learning is important for understanding the data from scratch. It involves defining the parameters, choosing the actions from the current state and also choosing the actions from the previous state and then developing a Q-table for maximizing the results or output rewards.

The 4 steps that are involved in Q-Learning:

  1. Initializing parameters – The RL (reinforcement learning) model learns the set of actions that the agent requires in the state, environment and time.
  2. Identifying current state – The model stores the prior records for optimal action definition for maximizing the results. For acting in the present state, the state needs to be identified and perform an action combination for it.
  3. Choosing the optimal action set and gaining the relevant experience – A Q-table is generated from the data with a set of specific states and actions, and the weight of this data is calculated for updating the Q-Table to the following step.
  4. Updating Q-table rewards and next state determination – After the relevant experience is gained and agents start getting environmental records. The reward amplitude helps to present the subsequent step.  

In case the Q-table size is huge, then the generation of the model is a time-consuming process. This situation requires Deep Q-learning.

Hopefully, this write-up has provided an outline of Deep Q-Learning and its related concepts. If you wish to learn more about such topics, then keep a tab on the blog section of the E2E Networks website.

Reference Links



This is a decorative image for: GAUDI: A Neural Architect for Immersive 3D Scene Generation
October 13, 2022

GAUDI: A Neural Architect for Immersive 3D Scene Generation

The evolution of artificial intelligence in the past decade has been staggering, and now the focus is shifting towards AI and ML systems to understand and generate 3D spaces. As a result, there has been extensive research on manipulating 3D generative models. In this regard, Apple’s AI and ML scientists have developed GAUDI, a method specifically for this job.

An introduction to GAUDI

The GAUDI 3D immersive technique founders named it after the famous architect Antoni Gaudi. This AI model takes the help of a camera pose decoder, which enables it to guess the possible camera angles of a scene. Hence, the decoder then makes it possible to predict the 3D canvas from almost every angle.

What does GAUDI do?

GAUDI can perform multiple functions –

  • The extensions of these generative models have a tremendous effect on ML and computer vision. Pragmatically, such models are highly useful. They are applied in model-based reinforcement learning and planning world models, SLAM is s, or 3D content creation.
  • Generative modelling for 3D objects has been used for generating scenes using graf, pigan, and gsn, which incorporate a GAN (Generative Adversarial Network). The generator codes radiance fields exclusively. Using the 3D space in the scene along with the camera pose generates the 3D image from that point. This point has a density scalar and RGB value for that specific point in 3D space. This can be done from a 2D camera view. It does this by imposing 3D datasets on those 2D shots. It isolates various objects and scenes and combines them to render a new scene altogether.
  • GAUDI also removes GANs pathologies like mode collapse and improved GAN.
  • GAUDI also uses this to train data on a canonical coordinate system. You can compare it by looking at the trajectory of the scenes.

How is GAUDI applied to the content?

The steps of application for GAUDI have been given below:

  • Each trajectory is created, which consists of a sequence of posed images (These images are from a 3D scene) encoded into a latent representation. This representation which has a radiance field or what we refer to as the 3D scene and the camera path is created in a disentangled way. The results are interpreted as free parameters. The problem is optimized by and formulation of a reconstruction objective.
  • This simple training process is then scaled to trajectories, thousands of them creating a large number of views. The model samples the radiance fields totally from the previous distribution that the model has learned.
  • The scenes are thus synthesized by interpolation within the hidden space.
  • The scaling of 3D scenes generates many scenes that contain thousands of images. During training, there is no issue related to canonical orientation or mode collapse.
  • A novel de-noising optimization technique is used to find hidden representations that collaborate in modelling the camera poses and the radiance field to create multiple datasets with state-of-the-art performance in generating 3D scenes by building a setup that uses images and text.

To conclude, GAUDI has more capabilities and can also be used for sampling various images and video datasets. Furthermore, this will make a foray into AR (augmented reality) and VR (virtual reality). With GAUDI in hand, the sky is only the limit in the field of media creation. So, if you enjoy reading about the latest development in the field of AI and ML, then keep a tab on the blog section of the E2E Networks website.

Reference Links




Build on the most powerful infrastructure cloud

A vector illustration of a tech city using latest cloud technologies & infrastructure