Peter Gregory

Automatic planning for machine tool calibration: A case study

Machine tool owners require knowledge of their machines capabilities, and the emphasis increases with areas of high accuracy manufacturing. An aspect of a machines capability is its geometric accuracy. International Standards and best-practice guides are available to aid understanding of the required measurements and to advise on how to perform them. However, there is an absence of any intelligent method capable of optimising the duration of a calibration plan, minimising machine down-time. In this work, artificial intelligence in the form of automated planning is applied to the problem of machine tool pseudo-static geometric error calibration. No prior knowledge of Artificial Intelligence (AI) planning is required throughout this paper. The authors have written this paper for calibration engineers to see the benefits that automated planning can provide. Two models are proposed; the first produces a sequential calibration plan capable of finding the optimal calibration plan. The second model has the additional possibility of planning for concurrent measurements, adding the possibility of further reducing machine down-time. Both models take input regarding a machines configuration and available instrumentation. The efficacy of both models is evaluated by performing a case study of a five-axis gantry machine, whereby calibration plans are produced and compared against both an academic and industrial expert. From this, the effectiveness of this novel method for producing optimal calibration plan is evaluated, stimulating potential for future work.

Constraint Based Planning with Composable Substate Graphs

Constraint satisfaction techniques provide powerful inference algorithms that can prune choices during search. Constraint-based approaches provide a useful complement to heuristic search optimal planners. We develop a constraint-based model for cost-optimal planning that uses global constraints to improve the inference in planning. n nThe key novelty in our approach is in a transformation of the SAS+ input that adds a form of macro-action to fully connect chains of composable operators. This translation leads to the development of a natural dominance constraint on the new problem which we add to our constraint model. n nWe provide empirical results to show that our planner, Constance, solves more instances than the current best constraint-based planners. We also demonstrate the power of our new dominance constraints in this representation.

A New Empirical Study of Weak Backdoors

Work by Kilby, Slaney, Thiebaux and Walsh [1] showed that the backdoors and backbones of unstructured Random 3SAT instances are largely disjoint. In this work we extend this study to the consideration of backdoors in SAT encodings of structuredproblems. We show that the results of Kilby et al.also apply to structured problems. Further, we analyse the frequency with which individual variables appear in backdoors for specific problem instances. In all problem classes there are variables with particularly high frequencies of backdoor membership. Backbone variables that do appear in backdoors typically appear in very few.

A meta-CSP model for optimal planning

One approach to optimal planning is to first start with a sub- optimal solution as a seed plan, and then iteratively search for shorter plans. This approach inevitably leads to an increase in the size of the model to be solved. We introduce a reformulation of the planning problem in which the problem is described as a meta-CSP, which controls the search of an underlying SAT solver. Our results show that this approach solves a greater number of problems than both Maxplan and Blackbox, and our analysis discusses the advantages and disadvantages of searching in the backwards direction.

The GRL System: Learning Board Game Rules with Piece-Move Interactions

Many real-world systems can be represented as formal state transition systems. The modeling process, in other words the process of constructing these systems, is a time-consuming and error-prone activity. In order to counter these difficulties, efforts have been made in various communities to learn the models from input data. One learning approach is to learn models from example transition sequences. Learning state transition systems from example transition sequences is helpful in many situations. For example, where no formal description of a transition system already exists, or when wishing to translate between different formalisms.

Automated planning for multi-objective machine tool calibration: optimising makespan and measurement uncertainty

The evolution in precision manufacturing has resulted in the requirement to produce and maintain more accurate machine tools. This new requirement coupled with desire to reduce machine tool downtime places emphasis on the calibration procedure during which the machines capabilities are assessed. Machine tool downtime is significant for manufacturers because the machine will be unavailable for manufacturing use, therefore wasting the manufacturers time and potentially increasing lead-times for clients. In addition to machine tool downtime, the uncertainty of measurement, due to the schedule of the calibration plan, has significant implications on tolerance conformance, resulting in an increased possibility of false acceptance and rejection of machined parts. n nThe work presented in this paper is focussed on expanding a developed temporal optimisation model to reduce the uncertainty of measurement. Encoding the knowledge in regular PDDL requires the discretization of non-linear, continuous temperature change and implementing the square root function. The implementation shows that not only can domain-independent automated planning reduce machine downtime by 10.6% and the uncertainty of measurement by 59%, it is also possible to optimise both metrics reaching a compromise that is on average 9% worse that the best-known solution for each individual metric.

A Monte-Carlo path planner for dynamic and partially observable environments

In this paper, we present a Monte-Carlo policy rollout technique (called MOCART-CGA) for path planning in dynamic and partially observable real-time environments such as Real-time Strategy games. The emphasis is put on fast action selection motivating the use of Monte-Carlo techniques in MOCART-CGA. Exploration of the space is guided by using corridors which direct simulations in the neighbourhood of the best found moves. MOCART-CGA limits how many times a particular state-action pair is explored to balance exploration of the neighbourhood of the state and exploitation of promising actions. MOCART-CGA is evaluated using four standard pathfinding benchmark maps, and over 1000 instances. The empirical results show that MOCART-CGA outperforms existing techniques, in terms of search time, in dynamic and partially observable environments. Experiments have also been performed in static (and partially observable) environments where MOCART-CGA still requires less time to search than its competitors, but typically finds lower quality plans.

Modelling road traffic incident management problems for automated planning

Abstract This paper is concerned with the application of automated planning in assisting the decision making and logistics in the area of Road Traffic Incidents. The characteristics of this area are that goals must be posed and plans must be output in real time. The domain is complex, with road topology, information distribution, traffic flows, driver behaviour, and highway management controls all potential factors. The representation and encoding of such domain knowledge, of possible actions and plans, and of potential tasks for the road traffic accident scenario is thus a crucial but difficult issue. The goal of this paper is to explore the potential of automated planning with hierarchical object-centred domain models in the application of Road Traffic Incident Management(RTIM).

OCL plus: processes and events in object-centred planning

An important area in AI Planning is the expressiveness of planning domain nspecification languages such as PDDL, and their aptitude for modelling real napplications. This paper presents OCLplus, an extension of a hierarchical object ncentred planning domain definition language, intended to support the representation nof domains with continuous change. The main extension in OCLplus provides nthe capability of interconnection between the planners and the changes that are ncaused by other objects of the world. To this extent, the concept of event and process nare introduced in the Hierarchical Task Network (HTN), object centred planning nframework in which a process is responsible for either continuous or discrete nchanges, and an event is triggered if its precondition is met. We evaluate the use of nOCLplus and compare it with a similar language, PDDL+.