This paper presents two decentralized algorithms for task allocation among a fleet of autonomous vehicles: the consensus-based auction algorithm (CBAA) and its extension to the multi-assignment problem, the consensus-based bundle algorithm (CBBA). These algorithms combine a market-based decision strategy for task selection with a consensus routine for conflict resolution. The algorithms are designed to handle inconsistencies in situational awareness (SA) and variations in communication network topology. The paper proves that both algorithms guarantee convergence to a conflict-free assignment under reasonable assumptions on the scoring scheme. Numerical experiments show that CBAA and CBBA outperform existing auction-based algorithms in terms of convergence properties and performance. The key difference from previous consensus-based methods is that the consensus routine is used to agree on winning bid values rather than SA. The CBAA is presented for the single-assignment problem, while CBBA extends it to the multi-assignment problem. The paper also discusses the convergence properties of CBBA, including its performance guarantees and robustness to inconsistent SA.This paper presents two decentralized algorithms for task allocation among a fleet of autonomous vehicles: the consensus-based auction algorithm (CBAA) and its extension to the multi-assignment problem, the consensus-based bundle algorithm (CBBA). These algorithms combine a market-based decision strategy for task selection with a consensus routine for conflict resolution. The algorithms are designed to handle inconsistencies in situational awareness (SA) and variations in communication network topology. The paper proves that both algorithms guarantee convergence to a conflict-free assignment under reasonable assumptions on the scoring scheme. Numerical experiments show that CBAA and CBBA outperform existing auction-based algorithms in terms of convergence properties and performance. The key difference from previous consensus-based methods is that the consensus routine is used to agree on winning bid values rather than SA. The CBAA is presented for the single-assignment problem, while CBBA extends it to the multi-assignment problem. The paper also discusses the convergence properties of CBBA, including its performance guarantees and robustness to inconsistent SA.