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

📌 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: Study: New York Times Has Published Extensive AI-Generate

The odds that the New York Times and other major news outlets have published AI-generated articles — whether knowing it or not — seem very high indeed.

Speculation abounded on this possibility earlier this week, centering on a “Modern Love” column published in the NYT last November. It was sparked when on X, Becky Tuch of Lit Mag News posted an excerpt of the piece with her controversial take: “this reads EXACTLY like AI slop,” she wrote.

Turns out there’s evidence that Tuch was onto something, a new piece in The Atlantic reveals.

The writer of the column, Kate Gilgan, told magazine that she hadn’t copy and pasted language from an AI model, but “did utilize AI as a tool,” including chatbots like ChatGPT, Claude, and Gemini, for seeking “inspiration and guidance and correction.”

“I used AI as a collaborative editor and not as a content generator,” Gilgan insisted.

At this point in the AI boom, when we know that AI’s effects on its users are often much farther-reaching than they realize, this feels like a thin distinction to make. In the process of constantly consulting a chatbot, it seems inevitable that its style and form could rub off on you.

And the scale may be imposing. Controversies like those surrounding Gilgan’s column inspired several AI researchers to go back and see how much AI material has crept into American newspapers.

Using an AI-detection tool from the startup Pangram Labs, their findings, published as a preprint study in October, should raise alarm. They suggest that nine percent of newly-published articles are either partially or fully AI-generated, mostly in smaller, local outlets.

But when they focused on opinion pieces in “newspapers of records” including, the New York Times, the Wall Street Journal, and the Washington Post, they found that these were over six times more likely to contain AI-generated content than articles that came out of their newsroom.

Now, a disclaimer: many AI detectors, especially free ones, are notoriously unreliable. (A screenshot of an AI detector flagging a passage from Mary Shelley’s “Frankenstein” as “100 percent AI generated” recently went viral and generated heaps of mockery.) False accusations are happening all the time. But, for what it’s worth, Pangram tends to be held up as among the most reliable out there, a sentiment borne out in head-to-head tests.

Moreover, it’s noteworthy that the AI detector singled out opinion pieces as being AI generated, rather than news articles. These are often penned by writers who aren’t professional journalists and don’t work within the organization, meaning there’s less oversight on how they’re being written. In other words, it makes sense that AI content would turn up in opinion sections — which are often used to platform all kinds of claims in need of a heavy reality check — lending the AI detector some credibility. No one can deny that loads of AI-generated dreck is polluting scientific journals, so why should news outlets be spared?

This comes as many news organizations become uncomfortably tangled with AI companies. The Washington Post launched an AI-generated podcast feature that creates summaries of the paper’s latest stories, along with a chatbot that fields reader questions. The New York Times uses AI to generate headlines. Bloomberg provides AI-generated summaries of its articles. A senior manager at the Associated Press recently told staffers that “resistance” to AI was “futile.”

Letting these tools anywhere near newsrooms, though, could be a slippery slope. Last month, a senior Ars Technica reporter was caught accidentally using AI-fabricated quotes in an article, forcing the publication to issue a retraction. The reporter claimed he didn’t use AI to write the article itself, but he used a chatbot to summarize his notes, as a result of which he accidentally included a quote the AI hallucinated. He was terminated after an investigation.

More on AI: Novel Pulled From Shelves After Author Is Accused of Using AI

The post Study: New York Times Has Published Extensive AI-Generated Articles appeared first on Futurism.

🔗 Sumber: futurism.com