This　paper　considers　the　vehicle　formation　problem　with　nonlinear　and　heterogeneous　dynamic　model　and　constraint　conditions.　A　predictive　spacing　strategy　(PSS)　is　put　forward,　which　utilized　the　present　velocity　information　of　the　vehicle　to　predict　the　distance　traveled　over　a　period　of　time.　A　distributed　model　predictive　control　(DMPC)　based　on　the　new　PSS　is　presented　for　the　heterogeneous　vehicle　formation　with　velocity,　acceleration　constraints　and　general　communication　topology.　A　sufficient　condition　to　ensure　the　asymptotic　stability　of　the　vehicle　formation　is　derived,　and　the　design　process　of　the　distributed　model　predictive　control　is　given.　Finally,　the　effectiveness　of　the　proposed　PSS　and　DMPC　algorithm　is　proved　by　experimental　simulation,　and　the　results　show　it　can　effectively　improve　the　effectiveness　of　vehicle　formation.　©　2023　IEEE.