| 摘要: | 電子商務隨著資訊科技的廣泛運用和行動網絡的蓬勃發展，已成為市場主流，帶動了「少量、多樣化」的市場需求趨勢。這種變化不僅加劇了物流中心的運營挑戰，特別是在揀貨作業方面更為顯著。根據 De Koster et al.（2007）研究，物流中心仍以人工作業為主，揀貨成本佔總運營成本的50%以上，影響整體效率與利潤。為提升競爭力，導入自動化設備與智能揀貨策略已成為降低成本、提高產能與提升運營效率的關鍵。此外，透過優化倉儲佈局、應用先進數據分析與機器學習技術，可進一步提升訂單處理速度與準確性，滿足即時配送需求，打造更靈活高效的供應鏈管理模式。
本研究利用類LocusBots系統的最大優勢，發揮其可靈活增減揀貨機器人的特性，使企業能有效應對訂單淡旺季的變化，提升倉儲管理的彈性。該系統運用動態路徑規劃技術，能即時更新揀貨環境狀態，智能閃避障礙物，以及最佳揀貨路徑計算，確保作業流暢且高效。此外，揀貨員無需在倉庫內長距離移動，只需留在指定揀貨區域，由機器人負責運送訂單和揀貨箱，使揀貨員專注於取貨，減少體力消耗與錯誤率。透過機器人協作運作，可大幅降低人力成本，提升揀貨效率與準確性，同時提高訂單處理速度，滿足即時配送需求，打造更智慧化的物流運營模式。
基於上述原因，本研究專注於類 LocusBots 系統中的揀貨策略。在物流揀貨倉庫中，倉庫被劃分為多個揀貨區域（Zones），每個區域包含相同數量的揀貨區塊（Blocks）。本研究分別探討「Robot的揀貨區域（Zone）選擇問題」與「Robot的揀貨區塊（Block）選擇問題」的各種法則。此外，本研究提出了兩種同時考量這兩個問題的揀貨流程。透過模擬與分析，本研究旨在找出具備最小化移動成本為核心邏輯的「SDZ搭配SDB法則」，在多數情境下對TST與TTIS皆具有穩定的改善效果，是本研究中最穩健且具普適性的高效能策略組合。
;With the widespread adoption of information technology and the rapid development of mobile networks, e-commerce has become a mainstream market force, driving a growing trend toward "small-volume, high-variety" demand. This shift has intensified the operational challenges faced by distribution centers, particularly in order picking operations. According to De Koster et al. (2007), manual operations still dominate warehouse activities, with picking costs accounting for over 50% of total operational expenses, thereby significantly impacting overall efficiency and profitability. To enhance competitiveness, the implementation of automated equipment and intelligent picking strategies has become critical for reducing costs, increasing productivity, and improving operational performance. Furthermore, optimizing warehouse layouts and leveraging advanced data analytics and machine learning techniques can further accelerate order processing and improve accuracy, enabling real-time fulfillment and supporting a more agile and efficient supply chain management model.
This study leverages the key advantage of LocusBot-like systems—their flexibility in scaling the number of picking robots—to help enterprises effectively respond to seasonal fluctuations in order volumes and enhance the adaptability of warehouse management. By employing dynamic path planning algorithms, the system can continuously update the picking environment in real time, intelligently avoid obstacles, and calculate optimal picking routes, thereby ensuring smooth and efficient operations. In this setup, pickers are no longer required to move extensively throughout the warehouse. Instead, they remain within designated picking zones while robots transport order containers, allowing pickers to focus solely on item retrieval, reducing physical fatigue and minimizing error rates. Through coordinated robot operations, the system significantly lowers labor costs, improves picking efficiency and accuracy, and accelerates order fulfillment to meet real-time delivery demands, fostering a more intelligent and responsive logistics operation model.
Based on the aforementioned reasons, this study focuses on picking strategies within a LocusBot-like system. In the warehouse picking environment, the warehouse is divided into multiple picking zones, each containing an equal number of picking blocks. This research investigates various decision rules for two core problems: the zone selection problem and the block selection problem for picking robots. In addition, two picking processes that simultaneously consider both problems are proposed. Through simulation and analysis, the study identifies the "SDZ combined with SDB" strategy—centered on minimizing travel cost—as the most robust and universally effective approach. It consistently demonstrates stable improvements in both Total System Time (TST) and Order Total Time In System (TTIS) across various scenarios, making it the most reliable high-performance strategy in this research. |
