The field of embodied agents is moving towards a deeper understanding of the impact of visual perception on decision-making stability. Recent research has highlighted the importance of systematic quantification of visual bias in robotic manipulation, with a focus on factor isolation and perceptual-fairness validation protocols. The development of new benchmarks, such as those for visual bias and home safety inspection, is enabling more robust evaluation of embodied agents and vision-language models. Noteworthy papers in this area include: RoboView-Bias, which proposes a benchmark for systematically quantifying visual bias in robotic manipulation and demonstrates a mitigation strategy that substantially reduces visual bias. HomeSafeBench, which introduces a benchmark for evaluating embodied vision-language models in free-exploration home safety inspection tasks and reveals significant limitations in current models.