ROBOMARKET
Stay Informed
Daily coverage from 5916 trusted sources — robotics, AI agents, and industry analysis for the GCC.
Last Updated: 08:00 PM (3m ago)
Auto-refresh every 2 hours
Large Language Models Show Metacognitive Sensitivity in Medical Reasoning
arXiv:2608.14552v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly evaluated and used in medicine, but clinical usefulness depends on answer accuracy and whether confidence tracks evidence quality and uncertainty. We developed a controlled, psychophysics-inspired clinical benchmark to test diagnostic choice and confidence behavior in a medical LLM. The benchmark focused on probable Alzheimer-type neurocognitive disorder (AT-NCD) versus depression-related cognitive impairment (DRCI). We generated 45 synthetic vignettes varying evidence strength, conflicting evidence, and missing information. Each vignette was presented under three prompt variants, yielding 135 trials. In a pilot run with gpt-4.1-nano, all trials produced valid structured outputs. Across forced-choice trials, diagnostic accuracy was 93.5%, mean confidence was 78.4%, and AUROC2 was 0.876. Confidence increased with evidence distance from the diagnostic boundary, decreased when information was mis
SpotlessGS: Relightable 3D Gaussian Splatting under Dynamic Illumination for Robotic Perception
arXiv:2608.14713v1 Announce Type: new Abstract: Robots operating in dark or poorly lit environments rely on onboard lights, which often produce uneven illumination that degrades downstream perception tasks. Prior approaches based on 2D image enhancement lack reliable supervision and fail to preserve multi-view geometric consistency. To address these limitations, we extend Dark Gaussian Splatting (DarkGS) toward a more accurate and flexible relightable 3D reconstruction framework. First, we eliminate the need for explicit light parameter calibration by jointly optimizing lighting parameters within the Gaussian Splatting framework. Second, we introduce a low-frequency illumination model based on spherical harmonics (SH) to capture spatially varying residual and ambient lighting effects. Third, we incorporate an MLP-based Bidirectional Reflectance Distribution Function (BRDF) to model non-Lambertian reflectance. Experiments on synthetic and real-world datasets demonstrate that our method
FLOPs vs Real Work: The Importance of Replication in AI Efficiency Assessment
arXiv:2608.14550v1 Announce Type: new Abstract: AI efficiency has recently taken the spotlight in both academy and industry due to massive model scales, high energy demands, and environmental costs. While reporting Floating Point Operations (FLOPs) is a traditional approach for assessing computational costs, the relationship between FLOPs and execution time is not straightforward, as layers with the same number of FLOPs may not have the same execution time because some operations are more easily parallelized than others. This paper sets out to replicate the original experiments from a study that proposed the $\alpha-FLOPs$ estimation formula to verify whether the results remain applicable on newer, more powerful hardware. During the replication process, we identify limitations in the replication materials provided by the original study, including a lack of specific dependency details and transparency regarding regression data. Our results validate the thesis that raw FLOPs alone are no
MISTac: A Vision-Based Tactile Sensor for Minimally Invasive Surgery
arXiv:2608.14772v1 Announce Type: new Abstract: Minimally invasive and robot-assisted surgery offer many advantages over traditional open surgery, but deprive surgeons of tactile feedback and the ability to palpate tissue with their fingers. To address this lack of tactile feedback, we introduce the MISTac, a high resolution vision-based tactile sensor specifically designed for palpation in MIS. The sensor has a replaceable sensor tip with a diameter of 8 mm which allows it to fit through the trocars used in minimally invasive surgery. Its modular 3D-printed case design allows the use of bulky off-the-shelf illumination and imaging hardware that can easily be exchanged and upgraded. The sensor has an optical resolution of 176.68 $\mu m$, a tactile resolution of 250 $\mu m$, and can resolve forces as little as 24.3 mN. An in vivo study with the sensor shows its usability in minimally invasive surgery. We trained a machine learning model with the tactile data collected in the trial on a
7 Kimi K3 Features That Make Every Other Model Feel Outdated
Developers launch new models every week, but most barely change how you work. Kimi K3 is different—not because of benchmark charts, but because of a few small API changes that fundamentally affect how you use it. The first is reasoning_effort, which defaults to maximum, alongside 131,072 max_completion_tokens. Ask K3 to rename a variable, and it [ ] The post 7 Kimi K3 Features That Make Every Other Model Feel Outdated appeared first on Analytics Vidhya.
Webwright: Why AI Web Agents Should Write Code, Not Click
For years, web agents have worked one click at a time—and often fallen apart on long tasks. Microsoft Research’s Webwright makes a different bet: give the model a terminal and let it write the program instead. On long-horizon tasks, the same GPT-5.4 model jumps from 33.5% to 60.1% success. And instead of leaving behind a click trace, it leaves something you can actually use again: a command-line tool. The post Webwright: Why AI Web Agents Should Write Code, Not Click appeared first on Towards Data Science.
Three Generations of Autoscaling — And Why Agentic Traffic Breaks All of Them
How autonomous agents broke two decades of capacity planning — and what to build instead The post Three Generations of Autoscaling — And Why Agentic Traffic Breaks All of Them appeared first on Towards Data Science.
How to Perform Effective Project Management with AI
Become a more productive software engineer with LLMs The post How to Perform Effective Project Management with AI appeared first on Towards Data Science.
Loop Engineering for RAG: The Small Loops Inside Each Step, the Big Loops Across the Pipeline
Enterprise Document Intelligence [Vol.1 #13bis] - The four bricks return useful results most of the time. Loop engineering is what the system does the rest of the time: when retrieval misses, when generation fails the schema, when the listing comes back incomplete, when an API call times out. Three control surfaces (trigger, termination, recovery) and one rule that separates a useful loop from a spinning one The post Loop Engineering for RAG: The Small Loops Inside Each Step, the Big Loops Across the Pipeline appeared first on Towards Data Science.
Don't Claim Benchmark-Oriented Optimization Improves General Coding Capability -- Diverse Evaluation Is Required
arXiv:2608.13566v1 Announce Type: new Abstract: Post-training papers, model cards, and blog posts often treat scores on a small set of coding benchmarks (e.g., SWE-bench and LiveCodeBench) as evidence of broad coding capability, both for research artifacts and user-facing systems. We argue that optimization for these benchmarks leads to measuring task-specific performance, creating a meaning gap between measured scores and claims of general coding ability. We examine this gap with a Django-based case study benchmark suite we create. Evaluating foundation models and checkpoints post-trained on SWE-bench trajectories, we find that benchmark rankings frequently fail to generalize. Post-trained checkpoints show little cross-task transfer, and SWE-bench optimization yields limited or no gains on our tasks or on LiveCodeBench. Similarly, fine-tuning on individual Django modalities fails to transfer. We conclude that a small number of benchmarks is insufficient for evaluating diverse models u
Depth-Aware Sensitivity Analysis of Mixture-of-Experts Models via Magnitude-Based Expert Masking
arXiv:2608.13565v1 Announce Type: new Abstract: Mixture-of-Experts (MoE) architectures scale large language models (LLMs) while preserving computational efficiency through sparse activation. Despite their widespread adoption, the relative importance of individual MoE layers remains insufficiently characterized, particularly for model compression. This paper presents a systematic layer-wise sensitivity analysis of the Qwen3.6-35B-A3B model (40 MoE layers, 256 experts per layer, top-8 routing) using magnitude-based expert masking on the XLCoST cross-lingual code translation benchmark. We conduct a multi-phase study spanning 100, 300, and 500 prompt evaluation scales across three H100 GPU servers. Our central finding is that layer sensitivity is strongly depth-dependent: early layers (0-9) and middle layers (10-29) are highly fragile to expert masking, while late layers (30-39), and especially very-late layers (35-39), tolerate aggressive masking of low-magnitude experts. Flat all-layer m
Inducing Reward-Free Judging Rubrics that Reduce Over-Crediting in Agent Evaluation
arXiv:2608.13564v1 Announce Type: new Abstract: Evaluating language-model agents at scale increasingly relies on a second language model as an automatic judge, because the gold signal, an executable environment reward, is expensive, slow, or unavailable at deployment time. Such a judge is a reward-free proxy whose value depends on whether it can be trusted, yet existing judges either hand-write the scoring rubric, as in G-Eval, or fine-tune the judge's weights, and both tend to credit fluent but unsuccessful trajectories as successes. We instead induce the text of an agent-judging rubric from a small set of ground-truth-labeled trajectories, grounding it in true outcomes. We present RubricForge, which evolves a judge rubric by reflective evolution against labeled trajectories to maximize agreement with the environment reward, freezes it, and applies it to held-out trajectories in one model call with no environment access. The optimized artifact is human-readable text, so every verdict