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

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

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 Update ai: NASA Forced to Shut Down Largest Library, Throw Invaluable

NASA’s budget is still an unfathomable mess. The government shutdown late last year once again delayed proceedings to determine the space agency’s future — but if it were up to the Trump administration, NASA’s science budget would be slashed in half, an “extinction-level” inflection point for US space exploration and science.

If Congress were to have its druthers, on the other hand, NASA’s budget would largely remain unchanged, securing the future of dozens of important missions — both ongoing and planned — that the White House is looking to place on the chopping block.

Trapped between the two, the fate of the space agency remains in the air. Congress passed a short-term resolution on November 12 that left the government until January 31 to ratify NASA’s budget. NASA’s recently sworn-in administrator and former SpaceX space tourist, Jared Isaacman, has yet to officially comment on the matter, though he’s made it clear that he’s aligned with the Trump administration’s tripling down on private industry-led space exploration.

As uncertainty and confusion prevail, though, the Trump administration has taken it upon itself to gut entire buildings at NASA’s iconic Goddard Space Flight Center (GFSC), which played a key role in the development of its groundbreaking James Webb and Hubble space telescopes, alongside countless other key missions.

This week, news emerged that the Trump administration is even shutting down the center’s library — NASA’s largest — and threatening to destroy a still undetermined number of books, documents, and journals in the process.

As the New York Times reports, many of these invaluable artifacts haven’t been digitized or made available elsewhere. While a NASA spokesperson told the newspaper that the agency will review what to keep and what to throw away over the next 60 days, it’s a sobering glimpse at a federal agency in crisis.

In a move that was described as a “consolidation, not a closure” by NASA press secretary Bethany Stevens, the Trump administration is looking to close 13 buildings and more than 100 labs across the GSFC campus by March.

Stevens claimed the latest moves are part of a master plan reorganization effort that was first devised in 2022, several years before president Donald Trump took office. Per the NYT, seven other NASA libraries have already been shuttered since 2022. Three of them were closed in 2025.

In November, NASA staffers raised concerns over word that over a dozen buildings on the GSFC campus were being emptied without notice. It’s not just books and important documents on the line; the staffers warned that highly specialized equipment was at risk of being thrown away like trash as well.

Lawmakers have been furious at the Trump administration’s handling of the situation.

“The Trump Administration has spent the last year attacking NASA Goddard and its workforce and threatening our efforts to explore space, deepen our understanding of Earth, and spur technological advancements that make our economy stronger and nation safer,” senator Chris Van Hollen (D-MA) told the NYT. “These reports of closures at Goddard are deeply concerning — I will continue to push back on any actions that impact Goddard’s critical mission.”

The GSFC library contains important documentation about our efforts to study the cosmos, dating back to the Apollo era over half a century ago.

Critics of the moves to gut the campus argue it would be reckless to abandon these documents.

“It’s not like we’re so much smarter now than we were in the past,” planetary scientist Dave Williams, who took NASA up on its offer for an early retirement last year, told the NYT. “It’s the same people, and they make the same kind of human errors. If you lose that history, you are going to make the same mistakes again.”

More on NASA: NASA Staff Horrified at Plan to Throw Out Incredibly Specialized Science Equipment Like Garbage

The post NASA Forced to Shut Down Largest Library, Throw Invaluable Materials in Trash appeared first on Futurism.

🔗 Sumber: futurism.com