MAROKO133 Update ai: World’s smallest programmable robots think, swim, and sense temperatu

📌 MAROKO133 Breaking ai: World’s smallest programmable robots think, swim, and sen

Robots have just shrunk to the size of microorganisms.

Researchers at the University of Pennsylvania have unveiled what they describe as the world’s smallest fully programmable, autonomous robots, sporting a brain developed at the University of Michigan.

These microscopic swimming machines can sense their surroundings, make decisions, and operate independently for months at a time.

Barely visible to the naked eye, each robot measures about 0.2 by 0.3 by 0.05 millimeters, placing it squarely at the scale of bacteria and single-celled organisms.

Despite their size, the robots can move in complex patterns, respond to temperature changes, and even coordinate their motion in groups.

What makes them especially striking is their cost and longevity.

Each robot costs roughly a penny to make, runs on light, and contains no moving parts, a design choice that makes it remarkably durable despite operating in fluid environments.

Together, the machines represent a long-awaited breakthrough in microscale robotics, a field that has struggled for decades to combine independent motion, sensing, and computing at extremely small sizes.

Robots at microbial scale

For years, electronics have steadily shrunk, but robots have lagged behind. Independent motion at the microscale has been particularly challenging, largely because water behaves very differently at tiny scales.

“We’ve made autonomous robots 10,000 times smaller,” said Marc Miskin, assistant professor in electrical and systems engineering at Penn and senior author of the research.

“That opens up an entirely new scale for programmable robots.”

Operating in water at this scale is less like swimming and more like pushing through thick syrup.

Instead of propellers or joints, the robots use an elegant workaround: they move the surrounding fluid itself.

Rather than pushing against water directly, the robots generate an electric field that nudges ions in the liquid.

Those ions then push on nearby water molecules, creating thrust that moves the robot forward.

This propulsion system has no moving parts, allowing the robots to swim for months and be transferred easily using a micropipette.

They can also travel in coordinated groups, moving together much like schools of fish.

A brain powered by light

The robots’ intelligence comes from ultra-miniaturized computers developed at the University of Michigan.

These tiny processors must run on just 75 nanowatts of power, about 100,000 times less than a smartwatch.

“We saw that Penn Engineering’s propulsion system and our tiny computers were just made for each other,” said David Blaauw, a senior author of the study.

To make this possible, the team had to radically redesign how programs are written and executed at the microscale.

“We had to totally rethink the computer program instructions, condensing what conventionally would require many instructions for propulsion control into a single, special instruction,” Blaauw said.

Most of each robot’s surface is covered by solar cells, which harvest light for power and also double as optical receivers.

Light pulses are used both to power the robots and to program them, with each robot carrying a unique identifier that allows it to receive individualized instructions.

The current generation is equipped with temperature sensors capable of detecting differences within a third of a degree Celsius.

The robots can move toward warmer areas or report temperature changes by wiggling, a behavior likened to the honeybee “waggle dance.”

“This is really just the first chapter,” Miskin said. “We’ve shown that you can put a brain, a sensor and a motor into something almost too small to see, and have it survive and work for months.”

Future versions could carry additional sensors, store more complex programs, or operate in harsher environments, potentially transforming medicine and microscale manufacturing, according to the journal Science Robotics.

🔗 Sumber: interestingengineering.com

📌 MAROKO133 Hot ai: Adobe Research Unlocking Long-Term Memory in Video World Model

Video world models, which predict future frames conditioned on actions, hold immense promise for artificial intelligence, enabling agents to plan and reason in dynamic environments. Recent advancements, particularly with video diffusion models, have shown impressive capabilities in generating realistic future sequences. However, a significant bottleneck remains: maintaining long-term memory. Current models struggle to remember events and states from far in the past due to the high computational cost associated with processing extended sequences using traditional attention layers. This limits their ability to perform complex tasks requiring sustained understanding of a scene.

A new paper, “Long-Context State-Space Video World Models” by researchers from Stanford University, Princeton University, and Adobe Research, proposes an innovative solution to this challenge. They introduce a novel architecture that leverages State-Space Models (SSMs) to extend temporal memory without sacrificing computational efficiency.

The core problem lies in the quadratic computational complexity of attention mechanisms with respect to sequence length. As the video context grows, the resources required for attention layers explode, making long-term memory impractical for real-world applications. This means that after a certain number of frames, the model effectively “forgets” earlier events, hindering its performance on tasks that demand long-range coherence or reasoning over extended periods.

The authors’ key insight is to leverage the inherent strengths of State-Space Models (SSMs) for causal sequence modeling. Unlike previous attempts that retrofitted SSMs for non-causal vision tasks, this work fully exploits their advantages in processing sequences efficiently.

The proposed Long-Context State-Space Video World Model (LSSVWM) incorporates several crucial design choices:

1. Block-wise SSM Scanning Scheme: This is central to their design. Instead of processing the entire video sequence with a single SSM scan, they employ a block-wise scheme. This strategically trades off some spatial consistency (within a block) for significantly extended temporal memory. By breaking down the long sequence into manageable blocks, they can maintain a compressed “state” that carries information across blocks, effectively extending the model’s memory horizon.
2. Dense Local Attention: To compensate for the potential loss of spatial coherence introduced by the block-wise SSM scanning, the model incorporates dense local attention. This ensures that consecutive frames within and across blocks maintain strong relationships, preserving the fine-grained details and consistency necessary for realistic video generation. This dual approach of global (SSM) and local (attention) processing allows them to achieve both long-term memory and local fidelity.

The paper also introduces two key training strategies to further improve long-context performance:

• Diffusion Forcing: This technique encourages the model to generate frames conditioned on a prefix of the input, effectively forcing it to learn to maintain consistency over longer durations. By sometimes not sampling a prefix and keeping all tokens noised, the training becomes equivalent to diffusion forcing, which is highlighted as a special case of long-context training where the prefix length is zero. This pushes the model to generate coherent sequences even from minimal initial context.
• Frame Local Attention: For faster training and sampling, the authors implemented a “frame local attention” mechanism. This utilizes FlexAttention to achieve significant speedups compared to a fully causal mask. By grouping frames into chunks (e.g., chunks of 5 with a frame window size of 10), frames within a chunk maintain bidirectionality while also attending to frames in the previous chunk. This allows for an effective receptive field while optimizing computational load.

The researchers evaluated their LSSVWM on challenging datasets, including Memory Maze and Minecraft, which are specifically designed to test long-term memory capabilities through spatial retrieval and reasoning tasks.

The experiments demonstrate that their approach substantially surpasses baselines in preserving long-range memory. Qualitative results, as shown in supplementary figures (e.g., S1, S2, S3), illustrate that LSSVWM can generate more coherent and accurate sequences over extended periods compared to models relying solely on causal attention or even Mamba2 without frame local attention. For instance, on reasoning tasks for the maze dataset, their model maintains better consistency and accuracy over long horizons. Similarly, for retrieval tasks, LSSVWM shows improved ability to recall and utilize information from distant past frames. Crucially, these improvements are achieved while maintaining practical inference speeds, making the models suitable for interactive applications.

The Paper Long-Context State-Space Video World Models is on arXiv

The post Adobe Research Unlocking Long-Term Memory in Video World Models with State-Space Models first appeared on Synced.

🔗 Sumber: syncedreview.com