MAROKO133 Breaking ai: Disney’s Robot Olaf Is a Straight Up Nightmare Edisi Jam 01:17

📌 MAROKO133 Update ai: Disney’s Robot Olaf Is a Straight Up Nightmare Edisi Jam 01

Disney’s robot of its character “Olaf” is incredibly impressive.

It’s also kind of terrifying.

This weekend, engineers at Disney Research Hub published a paper describing how they brought the beloved talking snowman from the “Frozen” films to life as a walking robot — and boy did they succeed.

In a video explaining their work, you can see footage of the robot Olaf, true to its size in the movies, roaming around a Disney park and the lab.

To describe it as uncanny would be harsh, since it does near-perfectly emulate pretty much everything about the character, down to its subtle waddle. But you are left with the distinct impression that what you’re seeing shouldn’t be physically possible; you wouldn’t expect someone with the eyes and proportions of an anime character to step in the real world, and it’s much the same for a 3D-animated snowman. And yet here it is, waltzing before your eyes. You may no longer want to build a snowman ever again, for fear of it coming alive like this one has.

All that aside, it’s a remarkable feat of engineering. You couldn’t get something more true to the character if you shrunk a person and stuck them in an Olaf-suit. You can yoink off the robot snowman’s carrot nose, and it’ll gasp and cackle. It’ll casually smile and wave at you as it ambles by.

The robot posed an unusual challenge to the engineers, who were tasked with building no mere humanoid. It has a huge head the size of its torso supported by a tiny neck, and no visible legs — just feet. 

Nonetheless, they prevailed. With the legs, for example, the team concealed most of the lower limbs inside its snowball torso, leaving just the feet showing. 

But simply having the robot take short little steps wasn’t enough. They meticulously fine-tuned its gait to mimic Olaf’s from the movie, and in particular ensured it walked with a heel-to-toe pattern to make it as fluid as possible.  

To fine tune the robot’s movements, the engineer trained it in a virtual environment using a technique known as reinforcement learning, in which the robot’s AI is rewarded based on it achieving a specific objective across thousands of simulations.

Perfectionists through and through, the team used the technique to reduce the noise of the bot’s steps. No loud plodding for our stout snowman; after implementing some “impact reduction,” the bot’s footfalls dropped from nearly 82 dB in volume to just 64 dB. Even neck temperature was a consideration: since the neck contained a host of small actuators controlling the head, it was at risk of overheating. With reinforcement learning, the bot’s AI system learned to slightly adjust its posture to prevent a thermal buildup, thereby preventing the illusion of animation from melting before our eyes.

More on robots: Robots Tear Up Stage as Backup Dancers

The post Disney’s Robot Olaf Is a Straight Up Nightmare appeared first on Futurism.

🔗 Sumber: futurism.com

📌 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