A.1 List of objects, containers, surfaces, and rooms in the apartment We list all the objects that are included in our experimental environment. We use the object rearrangement tasks for evaluation. The tasks are randomly sampled from different distributions. Simple: this task is to move one object in the house to the desired location. Novel Simple: this task is to move one object in the house to the desired location.

How to claim Verizon's $20 outage credit While Samsung's AI Vision and food tracking is a work in progress, it can still be genuinely useful. At their core, refrigerators are relatively simple devices. If you're the type of person to view every extra feature as a component that could potentially go wrong, basic iceboxes are probably the kind you go for. But for those on the other end of the spectrum, Samsung's latest Bespoke Refrigerators with AI inside have more bells and whistles than you might think possible -- including an optional 32-inch screen. The model we tested for this review came out in the second half of 2025 and will continue to be on sale throughout 2026. Hardware will remain the same, the only changes will come in the form of an OTA software update slated for later this year that will add support for Google Gemini, improved food recognition/labeling and more.

A Sharpa North robot uses a camera at CES 2026 in Las Vegas. A Sharpa North robot uses a camera at CES 2026 in Las Vegas. ESTLast modified on Fri 9 Jan 2026 10.00 EST This year will be filled with robots that can fold your laundry, pick up objects and climb stairs, fridges that you can command to open by voice, laptops with screens that can follow you around the room on motorised hinges and the reimagining of the BlackBerry phone. Those are the predictions from the annual CES tech show in Las Vegas that took place this week. The sprawling event aims to showcase cutting-edge technology developed by startups and big brands. Many of these fancy developments will be available to actually buy, moving from outlandish concepts to production devices, although some are still limited to costly prototypes.

Autonomous robots are increasingly deployed in dynamic and unstructured environments, where they must plan and execute complex tasks under uncertainty. Classical planning approaches, typically modeled in PDDL and solved with heuristic search, provide a principled foundation for task planning (Edelkamp and Schr odl, 2011; Geffner and Bonet, 2013). However, these methods rely on explicit domain models that enumerate preconditions and effects of actions. In practice, such models often omit implicit commonsense knowledge, for example, that a container must be upright before pouring, or that water must be boiled before making tea. The absence of such knowledge can lead to plans that are logically correct but physically invalid. Cognitive robotics research seeks to bridge symbolic reasoning with robot perception and control (Ghallab et al., 2004). While significant progress has been made in integrating planning with motion control and execution, robots still lack the ability to autonomously infer commonsense constraints that humans consider obvious. Large Language Models (LLMs), trained on massive corpora of human knowledge, present a promising avenue for addressing this gap. LLMs can generate likely preconditions, subgoals, and contextual constraints from natural language task descriptions, potentially enriching classical planning models. 1

What the Rocco fridge lacks in smarts, it makes up for in looks. You have a few hours left to save on the brand's Black Friday sale. When's the last time you poured a perfect glass of Pinot Noir in your own home? Red wines should be served somewhere between 58 and 68 degrees (opinions vary). That's a bit cooler than room temperature, but unless you want to dedicate money and space to a special refrigerator, you don't have many good options.

Ensuring the safety of embodied AI agents during task planning is critical for real-world deployment, especially in household environments where dangerous instructions pose significant risks. Existing methods often suffer from either high computational costs due to preference alignment training or over-rejection when using single-agent safety prompts. To address these limitations, we propose MADRA, a training-free Multi-Agent Debate Risk Assessment framework that leverages collective reasoning to enhance safety awareness without sacrificing task performance. MADRA employs multiple LLM-based agents to debate the safety of a given instruction, guided by a critical evaluator that scores responses based on logical soundness, risk identification, evidence quality, and clarity. Through iterative deliberation and consensus voting, MADRA significantly reduces false rejections while maintaining high sensitivity to dangerous tasks. Additionally, we introduce a hierarchical cognitive collaborative planning framework that integrates safety, memory, planning, and self-evolution mechanisms to improve task success rates through continuous learning. We also contribute SafeAware-VH, a benchmark dataset for safety-aware task planning in VirtualHome, containing 800 annotated instructions. Extensive experiments on AI2-THOR and VirtualHome demonstrate that our approach achieves over 90% rejection of unsafe tasks while ensuring that safe-task rejection is low, outperforming existing methods in both safety and execution efficiency. Our work provides a scalable, model-agnostic solution for building trustworthy embodied agents.

Language model (LM) pre-training is useful in many language processing tasks. But can pre-trained LMs be further leveraged for more general machine learning problems?