Connect with us

Tech

Robots trained with spatial dataset show improved object handling and awareness

Published

on

Robots trained with spatial dataset show improved object handling and awareness


Credit: CC0 Public Domain

When it comes to navigating their surroundings, machines have a natural disadvantage compared to humans. To help hone the visual perception abilities they need to understand the world, researchers have developed a novel training dataset for improving spatial awareness in robots.

In new research, experiments showed that robots trained with this dataset, called RoboSpatial, outperformed those trained with baseline models at the same robotic task, demonstrating a complex understanding of both spatial relationships and physical object manipulation.

For humans, shapes how we interact with the environment, from recognizing different people to maintaining an awareness of our body’s movements and position. Despite previous attempts to imbue robots with these skills, efforts have fallen short as most are trained on data that lacks sophisticated spatial understanding.

Because deep spatial comprehension is necessary for intuitive interactions, if left unaddressed, these spatial reasoning challenges could hinder future AI systems’ ability to comprehend complex instructions and operate in dynamic environments, said Luke Song, lead author of the study and a current Ph.D. student in engineering at The Ohio State University.

“To have true general-purpose foundation models, a robot needs to understand the 3D world around it,” he said. “So spatial understanding is one of the most crucial capabilities for it.”

The study was recently given as an oral presentation at the Conference on Computer Vision and Pattern Recognition. The work is published in the journal 2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).

To teach robots how to better interpret perspective, RoboSpatial includes more than a million real-world indoor and tabletop images, thousands of detailed 3D scans, and 3 million labels describing rich spatial information relevant to robotics. Using these vast resources, the framework pairs 2D egocentric images with full 3D scans of the same scene so the model learns to pinpoint objects using either flat-image recognition or 3D geometry.

According to the study, it’s a process that closely mimics visual cues in the real world.

For instance, while current training datasets might allow a robot to accurately describe a “bowl on the table,” the model would lack the ability to discern where on the table it actually is, where it should be placed to remain accessible, or how it might fit in with other objects. In contrast, RoboSpatial could rigorously test these spatial reasoning skills in practical robotic tasks, first by demonstrating object rearrangement and then by examining the models’ capacity to generalize to new spatial reasoning scenarios beyond their original training data.

“Not only does this mean improvements on individual actions like picking up and placing things, but also leads to robots interacting more naturally with humans,” said Song.

One of the systems the team tested this framework on was a Kinova Jaco , an assistive arm that helps people with disabilities connect with their environment.

During training, it was able to answer simple close-ended spatial questions like “Can the chair be placed in front of the table?” or “Is the mug to the left of the laptop?” correctly.

These promising results reveal that normalizing spatial context by improving robotic perception could lead to safer and more reliable AI systems, said Song.

While there are still many unanswered questions about AI development and training, the work concludes that RoboSpatial has the potential to serve as a foundation for broader applications in robotics, noting that more exciting spatial advancements will likely branch from it.

“I think we will see a lot of big improvements and cool capabilities for robots in the next five to ten years,” said Song.

Co-authors include Yu Su from Ohio State and Valts Blukis, Jonathan Tremblay, Stephen Tyree and Stan Birchfield from NVIDIA.

More information:
Chan Hee Song et al, RoboSpatial: Teaching Spatial Understanding to 2D and 3D Vision-Language Models for Robotics, 2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (2025). DOI: 10.1109/cvpr52734.2025.01470

Citation:
Robots trained with spatial dataset show improved object handling and awareness (2025, November 13)
retrieved 13 November 2025
from https://techxplore.com/news/2025-11-robots-spatial-dataset-awareness.html

This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no
part may be reproduced without the written permission. The content is provided for information purposes only.





Source link

Tech

Snag a Feature-Packed Gaming Headset for Under $100

Published

on

Snag a Feature-Packed Gaming Headset for Under 0


Looking for a wallet-friendly gaming headset with big feature support? The Corsair Void Wireless V2 is currently marked down to just $80 at both Best Buy and Amazon, a healthy $50 discount from its usual retail price. This lightweight yet capable gaming headset was already a great buy before the discount, with wide compatibility and a comfortable design built for long grinds.

It’s one of the more comfortable gaming headsets I’ve had the opportunity to review, thanks to a combination of its super lightweight build and breathable mesh ear cups, and it even fits my oversized noggin. Because there’s no active noise canceling, it has a much more open and natural sound profile, which is nice for anyone who needs to remain aware of their surroundings while deep in a round of Arc Raiders.

One of the big selling points is Dolby Atmos, a spatial audio implementation that’s fairly uncommon at this price point, and basically unheard of at the marked down price. It’s only supported in a handful of games, but even without it the headset has great spatial audio support that I found particularly good for games like Satisfactory, where it’s more of an immersive addition than a mechanical benefit.

Where a lot of headsets will lock you down to one or two consoles, the Corsair Void Wireless V2 is happy to work with a wide variety of systems, thanks to both Bluetooth and low-latency 2.4 GHz via the USB dongle. That means you can game on PC, PlayStation 4 and PlayStation 5, Nintendo Switch, and even iOS/Android for mobile gaming. With a claimed battery life of up to 70 hours in ideal conditions, you won’t need to worry about charging often, although I’m not sure the mesh ear cups would be great for a long flight.

While the Corsair Void Wireless V2 is featured in our best gaming headsets roundup, other headsets undercut it at the same price point. With the discount, I’m very happy to recommend the Corsair over some of the other picks, particularly if you have a bigger head or prefer a less isolating experience than what some of the other headsets provide.



Source link

Continue Reading

Tech

We Found the Best Travel Cameras You’ll Actually Use on Vacation

Published

on

We Found the Best Travel Cameras You’ll Actually Use on Vacation


If I were buying a camera today, this is the body I would get, travel or otherwise. The A7C R is one of the smallest full-frame, interchangeable-lens cameras on the market. This is why it’s our top pick for travelers in our guide to the Best Mirrorless Cameras. The 61-MP sensor offers amazing detail and very good dynamic range (14.7 EV). The super fast autofocus system is among the best you’ll find in mirrorless camera, and there’s great subject tracking as well, making it perfect for shooting fast-moving scenes on the go.

While Sony’s R series cameras are mainly for high resolution still images, the video specs here are solid enough for the casual video user, with support for 4K/60 fps video in full-frame mode (1.2X crop) or oversampled 4K/30 fps video. Both support 10-bit 4:2 color depth, various Log formats, and even 16-bit RAW output to an external recorder.

My main gripe about the A7C R is the same as it was in my initial review. The viewfinder is cramped and low-resolution (2.36 million dots). It’s not a deal breaker for me, but it’s something to keep in mind and good reason to rent a camera before you invest.

The big question with this camera is, which lens do you pair it with for travel? There’s a compelling argument to be made for the Sony FE 24-105-mm f/4 G ($1,398), which gives you everything from wide to portrait with a little bit of extra reach as well. Another great option if you like primes is the Sony 40-mm f2.5 G ($798), which makes for a compact kit, and 40-mm is a surprisingly great focal length for travel in my experience.

★ Alternative: At $2,198, Sony’s A7C II (note, no R) is a bit cheaper. It uses Sony’s smaller 33-MP sensor but is otherwise very close in size and capabilities, with considerably more video chops than the A7C R. If you want to make videos as well as stills, the A7C II is a better choice.



Source link

Continue Reading

Tech

Business leaders see AI risks and fraud outpacing ransomware, says WEF | Computer Weekly

Published

on

Business leaders see AI risks and fraud outpacing ransomware, says WEF | Computer Weekly


Midway through a decade that is coming to be defined by the runaway acceleration of technological change, the threat of ransomware attacks seems to be dropping down the agenda in boardrooms around the world, with C-suite executives more concerned about growing risks arising from artificial intelligence (AI) vulnerabilities, cyber-enabled fraud and phishing attacks, disruption to supply chains, and exploitation of software vulnerabilities.

This is according to the fifth annual World Economic Forum (WEF) Global cybersecurity outlook report, based on a survey of 804 participants from 92 countries, including 316 chief information security officers (CISOs), 105 CEOs and 123 other C-suite executives such as chief risk or technology officers, conducted between August and September 2025, as well as workshop discussions and short polls conducted around the forum’s Global Future Councils and Cybersecurity meeting.

A total of 87% of these respondents believed risks from AI increased in the past year, compared with 13% who were neutral on the subject. Approximately 77% saw risks from fraud and phishing on the rise; 66% talked about supply chain disruption; and 58% identified vulnerability exploitation as a growing threat.

However, when it came to ransomware, just 54% saw rising risk levels, compared with 39% who expressed a neutral opinion, while the remainder of the respondents, approximately 7%, said the risk from ransomware actually decreased in 2025.

“Cyber security risk in 2026 is accelerating, fuelled by advances in AI, deepening geopolitical fragmentation and the complexity of supply chains,” wrote WEF managing director Jeremy Jurgens in the report’s preamble.

“These shifts are compounded by the enduring sovereignty dilemma and widespread cyber inequity, two factors that expose systemic vulnerabilities. The result is a threat environment where the speed and scale of attacks are testing the limits of traditional defences.”

AI risk factors

Digging deeper into some of the risk factors arising from AI, the C-suite said that data leaks, followed by advancing adversarial capabilities, were the most pressing concerns, followed by the technical security of AI systems, increasingly complex governance, legal risks around intellectual property and liability, and software supply chain and code development concerns.

Notably, the top two concerns swapped places in the 2026 report compared with last year – with 34% most concerned about data exposure this year compared with 22% in 2025, while the percentage of those most concerned about adversarial capabilities fell from 47% last year to 29% this year.

This likely reflects a changing, potentially maturing, attitude to AI risk, and the WEF said it was looking to a “turning point” in the AI risk landscape this year.

It said that even though the AI arms race between defenders and attackers shows no signs of slowing, attention is pivoting from “offensive innovation” towards less noisy – but arguably more dangerous – factors.

Some of the other data points in the report also appear to bear this out, with C-suite executives doubling down on structured processes and governance models to better manage AI.

Quoted in the report, Josephine Teo, Singapore’s minister for digital development and information and minister-in-charge of the country’s Cyber Security Agency and Smart Nation Group, said: “Developments in AI are reshaping multiple domains, including cyber security. Implemented well, these technologies can assist and support human operators in detecting, defending and responding to cyber threats.

“However, they can also pose serious risks such as data leaks, cyber attacks and online harms if they malfunction, or are misused.”

Teo urged a more forward-looking, practical and collaborative approach to the safe development and use of rapidly evolving tech such as AI.

“The risks transcend borders, and the challenge is to maximise AI’s benefits, including to strengthen our cyber resilience, while minimising its risks,” she said.

Ransomware still a live threat

However, despite the headline risks detailed in the WEF’s report, the ransomware threat has not gone away – as demonstrated by many of the most well-documented cyber attacks to have taken place in 2025, most of which were still ultimately driven by extortion.

Indeed, among those who identified as CISOs, ransomware remained the leading risk concern. While CEOs concern themselves more with broader business impacts of cyber crime, CISOs are understandably consumed by the operational disruption a successful ransomware attack can cause.

This may go some way to explaining the elevated concerns over cyber fraud revealed by the WEF’s data.

A total of 77% of respondents said they had seen an increase in cyber-enabled fraud and phishing, and 72% revealed that either they themselves or someone in their professional or personal networks had been affected by it – the most common forms of attack reported were phishing, payment fraud and identity theft.



Source link

Continue Reading

Trending