Evaluating Sequential Single-Item Auctions for Dynamic Task Allocation

摘要

Sequential single-item (SSI) auctions have proven a very effective technique for tackling static task allocation problems in multi-robot settings and are only recently being applied to dynamic task allocation problems. We complement existing work by evaluating the effects of using different auctioning and winner determination schemes when dealing with dynamically appearing tasks. To this end we investigate the use of plan modification versus re-planning and minimum cost and regret clearing for winner determination in the auction.