Pierre Dellacherie Algorithm
A representative evaluation-based algorithm among traditional Tetris AIs, which scores candidate placements using heuristic features such as landing height, number of cleared lines, and number of holes, and selects the placement with the highest total score.
Timeline (last 90 days)
Traditional Tetris AI already has mature solutions, such as evaluation functions based on the Pierre Dellacherie algorithm.
The Pierre Dellacherie algorithm does not perform deep search; instead, it scores the board state after each placement using heuristic features and selects the placement with the highest total score.
The heuristic features used by the Pierre Dellacherie algorithm include landing height, number of lines cleared, number of holes, row transitions, column transitions, well depth, and others.
The limitation of the Pierre Dellacherie algorithm is that the feature weights require manual tuning, and it assumes that the board becomes stationary immediately after a piece lands.
The sand collapse in Sandtris breaks the assumption that the board becomes stationary immediately, so Dellacherie-style evaluation can only serve as a starting point, and re-evaluation of the board after collapse or multi-step lookahead is also needed.
All Facts (5)
Traditional Tetris AI already has mature solutions, such as evaluation functions based on the Pierre Dellacherie algorithm.
50%UnverifiedThe Pierre Dellacherie algorithm does not perform deep search; instead, it scores the board state after each placement using heuristic features and selects the placement with the highest total score.
50%UnverifiedThe heuristic features used by the Pierre Dellacherie algorithm include landing height, number of lines cleared, number of holes, row transitions, column transitions, well depth, and others.
50%UnverifiedThe limitation of the Pierre Dellacherie algorithm is that the feature weights require manual tuning, and it assumes that the board becomes stationary immediately after a piece lands.
50%UnverifiedThe sand collapse in Sandtris breaks the assumption that the board becomes stationary immediately, so Dellacherie-style evaluation can only serve as a starting point, and re-evaluation of the board after collapse or multi-step lookahead is also needed.
50%