📌 TOPINDIATOURS Eksklusif ai: ByteDance Introduces Astra: A Dual-Model Architectur

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

📌 TOPINDIATOURS Hot ai: Researchers Just Discovered Something Startling About How

If you’ve been watching the intense partisan divide over the killing of Renee Good by an ICE agent in Minnesota this week, it’s likely felt as though different Americans are living in completely separate realities.

As it turns out, neuroscience might be able to explain why. In a new study whose findings will surprise absolutely no one who’s endured a fiery holiday dinner debate, researchers discovered that conservative and liberal brains don’t just arrive at fundamentally different conclusions, but take strikingly different paths to get there. It’s a fascinating piece of research which just might explain something about the yawning political divides currently tearing society apart.

The study, published in the journal PLOS One, explored why people rely on different types of evidence when explaining why things happen. Specifically, the paper’s authors at the University of Idaho sought to find why some people look for tons of high-level statistical information, while others are happy to listen to anecdotes or single authority figures to reach political conclusions.

In order to carry out the study, researchers recruited 583 adults from the US, who each took a political ideology survey as well as a test to measure their level of “cognitive reflection,” which measures the capacity to look past a knee-jerk response and engage in factual analysis.

Each participant was then given a fictional scenario about cash bail, a “pay-to-leave-jail” policy which critics say unjustly punishes the poor.

The participants were told that, out of the top 300 US cities by population, 100 had ended cash bail. They were then asked to evaluate whether the policy was “effective at reducing crime” based on 10 pieces of evidence. Each piece of information was either composed of statistical figures or testimony from political “experts” from groups like the Democratic Party, Republican Party, and the National Rifle Association, in order to measure which type of evidence each one gravitated toward.

Participants had the chance to go through as many bits of evidence as they wanted before they delivered their final analysis to the researchers — though not all took advantage of this. For example, the researchers found that the probability a participant would rely on a single data point for their conclusion rose from about 4 percent for “very liberal” people to over 37 percent for “very conservative” folks.

As the researchers concluded, those on the left “tend to be more likely to consult a comprehensive set of statistical data relative to those on the right.”

Outside of the left-right paradigm, those who scored high on the cognitive reflection test were more likely to compare and contrast all available statistical data to form their conclusions than those who scored lower.

“Importantly, our study shows that two major individual-level variables help to predict what type of ‘evidence seeker’ a given person is: whether or not they are ‘cognitively reflected’ and whether or not they are liberal/conservative,” the study’s lead author and professor of politics and philosophy at the University of Idaho Florian Justwan told PsyPost.

“Indeed, people’s political beliefs influence how they look for information (often without them realizing it),” Justwan continued.

Zooming out, the research adds to a body of work which suggests that left-leaning people have more trust in the scientific method than conservatives. While that isn’t exactly a bombshell revelation, it does paint an interesting picture of how people one both sides of the aisle come to their political conclusions.

Said another way: while the research surely won’t end the culture war bickering at Thanksgiving, it gives us a strong explanation why one person’s “obvious proof” is another person’s fake news — a fundamental rift which, if ever resolved, could completely change the political landscape.

More on brain studies: Social Media Is Absolutely Nuking Children’s Brains, New Research Finds

The post Researchers Just Discovered Something Startling About How Conservatives Pick Political Positions appeared first on Futurism.

🔗 Sumber: futurism.com