MAROKO133 Update ai: ByteDance Introduces Astra: A Dual-Model Architecture for Autonomous

📌 MAROKO133 Update ai: ByteDance Introduces Astra: A Dual-Model Architecture for A

The increasing integration of robots across various sectors, from industrial manufacturing to daily life, highlights a growing need for advanced navigation systems. However, contemporary robot navigation systems face significant challenges in diverse and complex indoor environments, exposing the limitations of traditional approaches. Addressing the fundamental questions of “Where am I?”, “Where am I going?”, and “How do I get there?”, ByteDance has developed Astra, an innovative dual-model architecture designed to overcome these traditional navigation bottlenecks and enable general-purpose mobile robots.

Traditional navigation systems typically consist of multiple, smaller, and often rule-based modules to handle the core challenges of target localization, self-localization, and path planning. Target localization involves understanding natural language or image cues to pinpoint a destination on a map. Self-localization requires a robot to determine its precise position within a map, especially challenging in repetitive environments like warehouses where traditional methods often rely on artificial landmarks (e.g., QR codes). Path planning further divides into global planning for rough route generation and local planning for real-time obstacle avoidance and reaching intermediate waypoints.

While foundation models have shown promise in integrating smaller models to tackle broader tasks, the optimal number of models and their effective integration for comprehensive navigation remained an open question.

ByteDance’s Astra, detailed in their paper “Astra: Toward General-Purpose Mobile Robots via Hierarchical Multimodal Learning” (website: https://astra-mobility.github.io/), addresses these limitations. Following the System 1/System 2 paradigm, Astra features two primary sub-models: Astra-Global and Astra-Local. Astra-Global handles low-frequency tasks like target and self-localization, while Astra-Local manages high-frequency tasks such as local path planning and odometry estimation. This architecture promises to revolutionize how robots navigate complex indoor spaces.

Astra-Global: The Intelligent Brain for Global Localization

Astra-Global serves as the intelligent core of the Astra architecture, responsible for critical low-frequency tasks: self-localization and target localization. It functions as a Multimodal Large Language Model (MLLM), adept at processing both visual and linguistic inputs to achieve precise global positioning within a map. Its strength lies in utilizing a hybrid topological-semantic graph as contextual input, allowing the model to accurately locate positions based on query images or text prompts.

The construction of this robust localization system begins with offline mapping. The research team developed an offline method to build a hybrid topological-semantic graph G=(V,E,L):

• V (Nodes): Keyframes, obtained by temporal downsampling of input video and SfM-estimated 6-Degrees-of-Freedom (DoF) camera poses, act as nodes encoding camera poses and landmark references.
• E (Edges): Undirected edges establish connectivity based on relative node poses, crucial for global path planning.
• L (Landmarks): Semantic landmark information is extracted by Astra-Global from visual data at each node, enriching the map’s semantic understanding. These landmarks store semantic attributes and are connected to multiple nodes via co-visibility relationships.

In practical localization, Astra-Global’s self-localization and target localization capabilities leverage a coarse-to-fine two-stage process for visual-language localization. The coarse stage analyzes input images and localization prompts, detects landmarks, establishes correspondence with a pre-built landmark map, and filters candidates based on visual consistency. The fine stage then uses the query image and coarse output to sample reference map nodes from the offline map, comparing their visual and positional information to directly output the predicted pose.

For language-based target localization, the model interprets natural language instructions, identifies relevant landmarks using their functional descriptions within the map, and then leverages landmark-to-node association mechanisms to locate relevant nodes, retrieving target images and 6-DoF poses.

To empower Astra-Global with robust localization abilities, the team employed a meticulous training methodology. Using Qwen2.5-VL as the backbone, they combined Supervised Fine-Tuning (SFT) with Group Relative Policy Optimization (GRPO). SFT involved diverse datasets for various tasks, including coarse and fine localization, co-visibility detection, and motion trend estimation. In the GRPO phase, a rule-based reward function (including format, landmark extraction, map matching, and extra landmark rewards) was used to train for visual-language localization. Experiments showed GRPO significantly improved Astra-Global’s zero-shot generalization, achieving 99.9% localization accuracy in unseen home environments, surpassing SFT-only methods.

Astra-Local: The Intelligent Assistant for Local Planning

Astra-Local acts as the intelligent assistant for Astra’s high-frequency tasks, a multi-task network capable of efficiently generating local paths and accurately estimating odometry from sensor data. Its architecture comprises three core components: a 4D spatio-temporal encoder, a planning head, and an odometry head.

The 4D spatio-temporal encoder replaces traditional mobile stack perception and prediction modules. It begins with a 3D spatial encoder that processes N omnidirectional images through a Vision Transformer (ViT) and Lift-Splat-Shoot to convert 2D image features into 3D voxel features. This 3D encoder is trained using self-supervised learning via 3D volumetric differentiable neural rendering. The 4D spatio-temporal encoder then builds upon the 3D encoder, taking past voxel features and future timestamps as input to predict future voxel features through ResNet and DiT modules, providing current and future environmental representations for planning and odometry.

The planning head, based on pre-trained 4D features, robot speed, and task information, generates executable trajectories using Transformer-based flow matching. To prevent collisions, the planning head incorporates a masked ESDF loss (Euclidean Signed Distance Field). This loss calculates the ESDF of a 3D occupancy map and applies a 2D ground truth trajectory mask, significantly reducing collision rates. Experiments demonstrate its superior performance in collision rate and overall score on out-of-distribution (OOD) datasets compared to other methods.

The odometry head predicts the robot’s relative pose using current and past 4D features and additional sensor data (e.g., IMU, wheel data). It trains a Transformer model to fuse information from different sensors. Each sensor modality is processed by a specific tokenizer, combined with modality embeddings and temporal positional embeddi…

Konten dipersingkat otomatis.

🔗 Sumber: syncedreview.com

📌 MAROKO133 Breaking ai: Nanoparticle screen achieves highest visual clarity visib

Researchers in Sweden have developed a display technology with the smallest pixels ever made, capable of producing the highest resolution the human eye can perceive.

The breakthrough could redefine the future of virtual and augmented reality by creating visuals that look identical to real life.

Scientists from Chalmers University of Technology, the University of Gothenburg, and Uppsala University collaborated on the study.

Their innovation, called retina E-paper, uses nanoparticles to control how light scatters, achieving lifelike color reproduction that can be tuned electrically.

The clarity of a screen depends on the size and number of its pixels. But current technologies like micro-LEDs hit a limit when pixels shrink below one micrometer.

Retina E-paper breaks that barrier with pixels measuring just 560 nanometres, smaller than the wavelength of visible light.

Nanoparticles of tungsten oxide control each pixel’s optical behavior.

By varying their size and arrangement, the researchers can fine-tune how light reflects, creating red, green, and blue hues. A small voltage can “switch off” the pixels, turning them black.

“This means that each pixel roughly corresponds to a single photoreceptor in the eye, i.e. the nerve cells in the retina that convert light into biological signals.

Humans cannot perceive a higher resolution than this,” said Andreas Dahlin, Professor at Chalmers University of Technology.

The display area matches the size of a human pupil and achieves a resolution beyond 25,000 pixels per inch (ppi), which is about 150 times denser than most smartphone screens.

Reflective screen mimics nature

Unlike LED or OLED displays, retina E-paper does not emit its own light. Instead, it reflects ambient light, similar to how bird feathers shimmer with color.

This approach drastically reduces energy consumption and allows the screen to be placed very close to the eye.

To demonstrate the technology, the team recreated Gustav Klimt’s painting ‘The Kiss’ on a surface just 1.4 by 1.9 millimeters, roughly one four-thousandth the area of a smartphone display.

Despite the tiny size, the image retained impressive detail.

“The technology that we have developed can provide new ways to interact with information and the world around us.

It could expand creative possibilities, improve remote collaboration, and even accelerate scientific research,” said Kunli Xiong, Assistant Professor at Uppsala University and lead author of the study.

Towards immersive virtual worlds

Researchers believe retina E-paper could transform how humans experience digital environments.

Its lifelike color accuracy and ultra-high pixel density make it ideal for compact devices such as VR or AR headsets.

“This is a major step forward in the development of screens that can be shrunk to miniature size while improving quality and reducing energy consumption,” said Giovanni Volpe, Professor at the University of Gothenburg.

“The technology needs to be fine-tuned further, but we believe that retina E-paper will play a major role in its field and will eventually have impact on us all.”

By replicating the visual fidelity of reality, retina E-paper brings science closer to creating virtual worlds that the human eye cannot distinguish from the real one.

The study is published in the journal Nature.

🔗 Sumber: interestingengineering.com