US 12,111,162 B1
Techniques for reactive route planning
Phillip Oliver Kriett, Lynnwood, WA (US); Quico Pepijn Spaen, Menlo Park, CA (US); Georgios Patsakis, Sunnyvale, CA (US); Diwakar Tiwari, Newcastle, WA (US); Akhand Pratap Singh, Seattle, WA (US); Ivan Borges Oliveira, Bellevue, WA (US); Andrew V. Goldberg, Emerald Hills, CA (US); Philip Mark Kaminsky, El Cerrito, CA (US); Karthik Ramakrishnan, Issaquah, WA (US); and Manik Kumar, Woodinville, WA (US)
Assigned to Amazon Technologies, Inc., Seattle, WA (US)
Filed by Amazon Technologies, Inc., Seattle, WA (US)
Filed on Jun. 21, 2022, as Appl. No. 17/845,439.
Int. Cl. G01C 21/34 (2006.01)
CPC G01C 21/3415 (2013.01) [G01C 21/3446 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A computer-implemented method, comprising:
obtaining, by a computing device, a candidate column pool, each candidate column of the candidate column pool modeling a route associated with one or more legs of a set of legs, each leg being associated with a start time, an origin location, and a destination location, and the candidate column pool being previously generated by a first column generation module utilizing the set of legs as input;
obtaining an existing route plan, the existing route plan associating a plurality of blocks with a subset of columns selected from the candidate column pool, a block of the plurality of blocks being associated with a block start time, a block end time, and a block state;
identifying one or more canceled legs and one or more locked legs of the set of legs;
generating a filtered route plan and a filtered candidate column pool based at least in part on at least one of the one or more canceled legs and the one or more locked legs;
generating a candidate set of legs based at least in part on identifying a newly-added leg, the candidate set of legs comprising the newly-added leg and at least one leg identified from the set of legs, and the candidate set of legs excluding the one or more locked legs and the one or more canceled legs;
generating an updated candidate column pool based at least in part on providing the filtered route plan, the filtered candidate column pool, and the candidate set of legs as input to a second column generation module;
generating an optimized route plan based at least in part on the updated candidate column pool and the plurality of blocks, the optimized route plan including at least one block and corresponding column of the existing route plan is unchanged in the optimized route plan; and
instructing one or more resources of a network according to the optimized route plan.