In　order　to　improve　driving　safety　in　blind-spot　dangerous　scenarios　of　which　pedestrians　traverse　in　front　of　stationary　obstacle　vehicle　(OV)　and　collide　with　the　host　vehicle　(HV),　a　pedestrian　collision　avoidance　system　based　on　vehicle-to-vehicle　(V2V)　communication　is　proposed.　Position　and　speed　information　of　the　pedestrian　is　detected　by　the　OV　sensors　and　transmitted　to　the　HV　by　V2V　communication,　and　the　relative　position　and　speed　information　between　HV　and　pedestrian　is　derived　by　coordinate　transformation.　Longitudinal　distance　model　including　warning　safety　distance,　braking-trigger　safety　distance,　and　minimum　safety　distance,　along　with　lateral　model　for　categorizing　danger　levels　are　established　to　determine　the　decision-making　actions　(normal　driving,　braking　or　steering)　of　HV.　A　multi-constraint　optimization-based　obstacle　avoidance　trajectory　planning　method　based　on　a　quintic　polynomial　is　presented　for　steering　scenario.　The　effectiveness　of　V2V-based　collision　avoidance　strategy　in　the　blind　spot　dangerous　scenario　is　simulated　and　verified　by　the　co-simulation　of　PreScan,　CarSim,　and　MATLAB.　The　results　reveal　that　through　the　utilization　of　V2V　communication,　HV　can　proactively　acquire　real-time　pedestrian　motion　data　to　determine　collision　avoidance　strategies　appropriately　under　different　emergency　scenarios,　and　thereby　ensuring　the　safety　of　both　AV　and　pedestrian.　©　2024　SPIE.