In last mile delivery logistics, the high uncertainty of the urban environment arises from traffic congestion, limited parking availability, and handover uncertainty that are found to cause significant delays and inefficiencies. To assist addressing these challenges, a parcel reshuffling algorithm is proposed, that aligns with the principles of the Physical Internet and can be deployed as a collaborative tool. An independent platform receives delivery status updates, from one or more operators, that are analyzed to identify collaboration options between vans dynamically. The proposed algorithm functions in four steps to: (1) identify suitable help candidate vehicles from the delivery rounds operating in proximity, (2) once the optimal help van is identified, reshuffle the parcels to assist the late running van, (3) identify a meeting point for exchanging the parcels, and (4) reroute both vans through the meeting point, and all the pending delivery locations. The algorithm is applied in a case study, in central Madrid, Spain utilizing a dataset provided by EU funded research project PLANET.