📌 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 Update ai: World’s first space delivery vehicle can bring supplies

Inversion, a young aerospace and defense company based in Los Angeles, has unveiled its first flagship spacecraft, Arc.

The reentry vehicle is designed to deliver up to 500 pounds of mission-critical cargo from orbit to almost any point on Earth in less than an hour. The company revealed the spacecraft during an event at its factory.

Co-founders Justin Fiaschetti and Austin Briggs, who started the company in 2021, presented Arc as a new kind of logistics platform.

“Arc represents the next leap, creating a logistics network in space that will make Earth radically more accessible,” Fiaschetti said.

Spacecraft designed for rapid drops

Arc stands about 8 feet tall and 4 feet wide, roughly the size of a large tabletop. It is built to handle deliveries ranging from medical kits to drones.

In an interview with Ars Technica, Fiaschetti said the company plans to keep Arcs in orbit for extended periods, ready to descend when called upon. “The nominal mission for us is pre-positioning Arcs on orbit, and having them stay up there for up to five years… being able to bring their cargo or effects to the desired location in under an hour,” he said.

The spacecraft is a lifting body design, meaning it can maneuver as it reenters the atmosphere.

According to the company, Arc has a cross-range of about 621 miles during reentry, allowing it to steer across wide areas before descent.

Instead of needing a runway, the vehicle lands under parachutes. Its propulsion system uses non-toxic materials, which allows soldiers to handle it safely without protective gear immediately after touchdown.

“We like to describe this as mission-enabling cargo or effects,” said Fiaschetti. “This could be a wide variety of specific payloads, anything from medical supplies to drones to what have you. But the key discriminator is, does this make a difference in the moment it’s needed when it gets back down to the ground?”

Hypersonic testing role

Beyond delivery, Inversion is pitching Arc as a hypersonic test platform.

The spacecraft can reach speeds above Mach 20, maintain extreme conditions for longer durations, and sustain heavy g-forces. U.S. defense agencies have increased funding and focus on hypersonic research, and Inversion believes Arc offers a cost-effective way to support that work.

“Fully reusable and capable of precise landings for rapid recovery, Arc makes hypersonic testing faster, repeatable, and more affordable,” the company said in its announcement. Inversion’s selection to participate in the Kratos-led MACH-TB 2.0 program indicates growing interest in Arc’s role as part of national testing infrastructure.

By combining maneuverability with reusability, Inversion argues that Arc provides both defense logistics and advanced research capabilities in one platform.

Building on earlier work

Arc follows Inversion’s smaller demonstration spacecraft, Ray, which launched in January on SpaceX’s Transporter-12 rideshare mission.

Ray weighed about 200 pounds and tested systems such as propulsion, avionics, and solar power. While Ray successfully adjusted its orbit and continues to function, it was not designed to land.

“Ray won’t be coming back,” Fiaschetti told Ars Technica. “We’re doing long-term testing of software on orbit.”

That test gave Inversion confidence to begin work on Arc. The company says it has already built a full-scale development unit of the primary structure, run dozens of drop tests, and completed aerodynamic modeling.

“Every milestone brings Arc closer to flight maturity, and the pace of progress is only accelerating,” said Briggs, the company’s chief technology officer.

The team has also partnered with NASA on a thermal protection system for extreme reentry conditions. Inversion, now 60 employees strong, aims to fly Arc’s first mission by 2026.

🔗 Sumber: interestingengineering.com