This thesis by David Chapman, submitted to the Massachusetts Institute of Technology (MIT) in 1985, focuses on the problem of achieving conjunctive goals in domain-independent planning. The author presents a simple and precise algorithm called TWEAK, which is a rational reconstruction of previous non-linear planners like NONLIN and NOAH. TWEAK is designed to be clear and rigorous, addressing the limitations of earlier planners that were often heuristic and ill-defined. The thesis begins with an introduction to conjunctive planning, explaining the challenges and the Sussman anomaly, a classic problem illustrating the difficulties in achieving multiple goals simultaneously. Chapman then details the structure of TWEAK, including its plan representation, goal achievement procedure, and top-level control structure. The plan representation uses constraint posting to specify partial orders on steps, allowing for flexible plan construction. The goal achievement procedure ensures that a plan necessarily solves the given problem, while the top-level control structure uses dependency-directed breadth-first search to explore alternate paths to achieve goals. Chapman also analyzes past and future planning research, highlighting the historical development of conjunctive planners and the limitations of current action representations. He suggests that future research should focus on extending understood techniques to more expressive action representations, addressing the "frame problem" and enabling real-world planning. The thesis concludes with a discussion of the correctness and completeness of TWEAK, proving that it will find a solution if one exists and will signal failure if no solution is found.This thesis by David Chapman, submitted to the Massachusetts Institute of Technology (MIT) in 1985, focuses on the problem of achieving conjunctive goals in domain-independent planning. The author presents a simple and precise algorithm called TWEAK, which is a rational reconstruction of previous non-linear planners like NONLIN and NOAH. TWEAK is designed to be clear and rigorous, addressing the limitations of earlier planners that were often heuristic and ill-defined. The thesis begins with an introduction to conjunctive planning, explaining the challenges and the Sussman anomaly, a classic problem illustrating the difficulties in achieving multiple goals simultaneously. Chapman then details the structure of TWEAK, including its plan representation, goal achievement procedure, and top-level control structure. The plan representation uses constraint posting to specify partial orders on steps, allowing for flexible plan construction. The goal achievement procedure ensures that a plan necessarily solves the given problem, while the top-level control structure uses dependency-directed breadth-first search to explore alternate paths to achieve goals. Chapman also analyzes past and future planning research, highlighting the historical development of conjunctive planners and the limitations of current action representations. He suggests that future research should focus on extending understood techniques to more expressive action representations, addressing the "frame problem" and enabling real-world planning. The thesis concludes with a discussion of the correctness and completeness of TWEAK, proving that it will find a solution if one exists and will signal failure if no solution is found.