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

📌 MAROKO133 Eksklusif ai: ByteDance Introduces Astra: A Dual-Model Architecture fo

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: Video: Genesis’ smart robotic brain enables human-level m

Genesis AI has unveiled GENE-26.5, a new robotic brain designed to give robots human-level physical manipulation abilities.

The company also introduced a scalable training system aimed at solving the long-standing data shortage in robotics foundation models.

The platform combines a proprietary human-scale dexterous robotic hand, which allows direct transfer of human skills to robots, with a new high-capacity data engine.

Genesis says the system can generate unlimited training data, enabling the development of more capable and productive general-purpose robots for real-world applications.

Humanlike robot dexterity

The robotics-focused AI foundation model is designed to process massive amounts of data and operate across different environments.

The system is built to help robots perform complex, long-duration tasks with advanced dexterity and human-like precision. According to the company, GENE-26.5 is designed to enable robots to quickly adapt to unfamiliar environments and new tasks without extensive retraining.

To demonstrate the model’s capabilities, Genesis released a video showing robots completing some of the most advanced manipulation tasks achieved so far. The demonstrations highlight smooth hand coordination, precise movements, and human-like control across a variety of activities.

In one sequence, the robot prepares a 20-step meal that includes chopping tomatoes, cracking eggs with one hand, and coordinating both hands during cooking. Another demonstration shows the robot making a smoothie by handling ingredients, pouring liquids, blending, and serving the drink mid-air using two-hand coordination.

The system also performs delicate laboratory tasks such as pipetting and liquid transfer, wire harnessing for electronics assembly, solving a Rubik’s Cube through continuous in-air manipulation, sorting multiple objects with one hand, and playing a fast-paced piano composition, according to a statement by Genesis.

Genesis claims the demonstrations prove that GENE-26.5 can give robots highly advanced physical manipulation skills that were previously impossible.

Robotic hands evolve

Genesis has also developed a proprietary robotic hand and data collection system designed to reduce the “embodiment gap” between humans and robots. The company says differences between human and robotic hand structures have long limited robots’ ability to learn effectively from human-generated data.

The new robotic hand closely matches the form and movement of a human hand and works alongside a wearable glove fitted with tactile-sensing electronic skin. The system creates a direct 1:1:1 mapping between the human hand, the glove, and the robotic hand, allowing human actions to be accurately transferred into robotic training data.

According to Genesis, the glove system is 100 times cheaper than conventional hardware solutions. Internal testing also showed up to five times greater data collection efficiency and higher-quality results compared to traditional teleoperation systems.

The company plans to deploy the gloves in real-world workplaces through industry partnerships. Workers can wear the gloves while performing routine tasks, allowing robots to learn directly from everyday human activities. Genesis AI says this approach will help create a large-scale library of human skills for robotics training.

Beyond glove-based data, the company’s data engine also uses egocentric videos recorded from wearable cameras and publicly available human-centered internet videos. Genesis AI says combining these data sources improves robotic learning efficiency and enables robots to perform more advanced physical tasks.

🔗 Sumber: interestingengineering.com