MAROKO133 Eksklusif ai: World’s first five-ton eVTOL flies full transition with 932-mile r

📌 MAROKO133 Update ai: World’s first five-ton eVTOL flies full transition with 932

AutoFlight has unveiled Matrix, a five-ton electric vertical takeoff and landing aircraft, and completed a public full-transition flight at its low-altitude test center.

The demonstration marked the first time an eVTOL in the five-ton class has flown from vertical takeoff to forward cruise and back to vertical landing in a single sequence.

The company says the flight validates its work on aerodynamics, high-power electric propulsion, and flight control software.

The aircraft transitioned cleanly between modes, a step developers see as critical for scaling eVTOLs beyond short hops.

Matrix targets heavier payloads and longer routes than most current designs.

Full-transition flight milestone

During the demonstration, Matrix lifted vertically, accelerated into wing-borne cruise, and then returned to a controlled vertical landing.

The company described the sequence as a full-mode transition flight. Engineers tracked stability and control across each phase.

Matrix measures 20 meters in wingspan, 17.1 meters in length, and 3.3 meters in height.

It flies with a maximum takeoff mass of 5,700 kg (12,566 lb). Developers designed the platform to support both passenger and cargo missions.

The passenger version supports flexible layouts. Operators can configure ten business-class seats or six VIP seats. The cargo variant targets heavier logistics.

It uses a hybrid power system and supports a maximum payload of 1,500 kg (3,307 lb).

A large forward-opening door fits two standard AKE air freight containers, a feature aimed at improving turnaround times for one-tonne-scale freight operations.

Design, safety, and range

Matrix uses a lift-and-cruise compound wing design with a triplane layout and a six-arm structure.

The configuration aims to maintain aerodynamic stability across hover, transition, and cruise.

The company says the architecture supports predictable handling during complex flight phases.

The pure electric version delivers a maximum range of 250 km (155 miles). A hybrid-electric variant extends range to 1,500 km (932 miles).

These figures position the platform for regional travel, heavy logistics, and emergency response missions that exceed typical urban air taxi distances.

Engineers built the platform to scale across mission profiles. The company emphasizes reliability, redundancy, and certification readiness as development priorities.

Matrix continues the company’s broader product roadmap. Earlier platforms include Great White Shark for industrial roles, CarryAll for autonomous logistics, and Prosperity for urban air mobility.

The Matrix program draws on experience in low-altitude operations, safety systems, and airworthiness certification.

AutoFlight CEO and founder Tian Yu framed the aircraft as a shift in market expectations.

“Matrix is not only a rising star in the aviation industry, but also an ambitious disruptor. It will eliminate the industry perception that eVTOL = short-haul, low payload, and reshape the rules of eVTOL routes.”

He added context on costs and scale, saying the aircraft reduces transportation costs per seat-kilometer and per ton-kilometer through economies of scale.

“It covers all scenarios, from urban travel to intercity connections, and fosters the expansion of the entire low-altitude ecosystem.”

The company plans to advance testing as it moves toward commercial readiness, targeting missions where payload and range matter as much as vertical access.

🔗 Sumber: interestingengineering.com

📌 MAROKO133 Update ai: Adobe Research Unlocking Long-Term Memory in Video World Mo

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