With the rapid growth of e-commerce and ongoing innovations in the logistics industry, intelligent unmanned logistics warehousing systems have emerged to significantly enhance operational efficiency and reduce costs. In these systems, the two critical stages of order assignment and path planning are interconnected through racks in the order picking process. However, prior research has largely overlooked their joint optimization. In this paper, we investigate the real time task planning problem (RTTP) in intelligent unmanned logistics warehousing, where racks are dynamically assigned to orders arriving in real time, and robots are responsible for delivering the racks to workstations according to planned paths, with the goal of jointly minimizing total order processing time and travel costs. To solve the RTTP, we first design a joint optimization evaluation indicator and propose a joint optimization task planning (JOTP) algorithm. Furthermore, we innovatively introduce a reinforcement learning-based approach (JOTP-RL) by modeling order selection as a partially observable Markov decision process (POMDP), and employing the Q-Mix algorithm to solve it. To enhance path planning efficiency, we optimize the improved THA\(^*\) algorithm by eliminating redundant calculations and accounting for congestion times. Finally, extensive experiments conducted on two datasets demonstrate that our proposed algorithms significantly outperform the baseline and state-of-the-art methods, achieving superior efficiency and effectiveness in both execution time and task optimization.

In Visual Question Answering (VQA), the model’s ability to understand and reason across different modalities—language, visual, and multimodal—is crucial for accurate predictions. However, recent studies have identified a significant challenge of language bias in VQA models, where the model’s reasoning often hinges on incorrect linguistic associations rather than genuine multimodal understanding. To addressthis issue, We propose a novel Ensemble Bias Gradient Debiasing Approach (EBGDA) that combines bias detection with a dynamic forgetting mechanism. Our method utilizes a bias detector to identify and score biases across linguistic, visual, and multimodal data, enabling the model to focus on unbiased informationfor more accurate predictions. Additionally, inspired by human reasoning, we introduce the Forgotten Attention Algorithm (FAA), which iteratively “forgets” irrelevant visual content, progressively concentrating attention on the image regions most relevant to the question. This combination of bias mitigation andattention focusing enhances the model’s ability to make multimodal inferences, reducing bias and improving overall performance. Extensive experiments on the VQA-CP v2, VQA v2, VQA-VS, GQA-OOD, and VQA-CE datasets demonstrate the effectiveness of our approach, showing state-of-the-art performance inmitigating biases and excelling in complex multimodal scenarios. Our approach achieves a 21.32% improvement over UpDn on VQA-CP v2, establishing a new state-of-the-art among methods without data augmentation.