Organizations: Cognitive Robotics, Delft University of Technology, Delft, The Netherlands · Robotron Datenbank Software GmbH, Dresden, Germany · Intelligent Robotics Lab, Universidad Rey Juan Carlos, Fuenlabrada, Madrid, Spain · Department of Informatics, University of Oslo, Oslo, Norway
Autonomous robots may need to adapt both their task plans and architectural configurations as operating conditions change. These decisions are interdependent: architectural configurations determine which actions a robot can perform, while task plans determine which capabilities are required. This paper presents PLANTA, an ontology-mediated approach to task and architecture co-adaptation that considers these dependencies during mission planning. PLANTA uses an ontology expressed in the Web Ontology Language to represent architectural alternatives and their associated functional and quality requirements, while the Planning Domain Definition Language defines task actions and mission goals. By automatically translating and integrating these models, PLANTA enables joint planning of task and reconfiguration actions. The approach uses satisficing planning and replans when changes during execution invalidate the current plan. PLANTA is evaluated in two robotic scenarios involving underwater inspection and ground navigation. The results support the feasibility of runtime co-adaptation and show lower planning time and memory consumption compared with optimal-policy generation in the evaluated navigation scenario, while producing plans of similar quality.
Figures & tables
Figure 1. PLANTA framework, organized into design time, deployment time, and runtime, separated by dashed vertical lines. Some PlanSys2 nodes are omitted for clarity, and the Planner and Plan Execution components shown were extended as part of this work. Left-to-right diagram of the \acronymworkflow. At design time, a PDDL domain and problem and a \pddltomasysontology are specified. At deployment time, \owltopddlcombines the ontology TBox with the domain and the ABox with the problem. The TACA transformation then modifies the domain. At runtime, the resulting domain and problem populate the \plansysPDDL knowledge base, which supplies the planner and plan execution component. The planner sends a plan to plan execution, which sends task and reconfiguration actions to the robotic system. Monitoring feeds observations from the robot back to the knowledge base. Solid arrows denote data flow and dashed arrows denote ROS interfaces.
Concept
Concept
Example
Example
OWL Ontology
PDDL
OWL Ontology
PDDL
Individual
Object
f1
(f1 - object)
Class
Unary predicate
\FunctionTomasys
(Function ?f)
Class Assertion
Grounded unary predicate
\FunctionTomasys(f1)
(Function f1)
Relation
Binary predicate
\functionGrounding
(functionGrounding ?f ?fd)
Relation Assertion
Grounded binary predicate
\functionGrounding(f1,fd1)
(functionGrounding f1 fd1)
Table 1. Mapping between OWL ontology and PDDL concepts following the approach of ( Borgwardt et al., 2022 ) .
Figure 4. Transformation of a SWRL rule into its equivalent PDDL derived predicate. Two side-by-side representations of the same rule. On the left, a SWRL implication states that if f is a Function and fd_unground is a FunctionDesign, then fd_unground solves f. On the right, a PDDL derived predicate for solvesF expresses the same conditions and uses an equality test to bind its first parameter to the constant fd_unground.
Figure 6. Derived predicates resolution graphs for the PDDL formulation in Figure 5 : (a) the full dependency graph and (b) the pruned graph for the plan π=(action1,action2) . The predicates predicateAA and predicateC are omitted to simplify the visualization, as they are not relevant to the graph structure due to having no child predicates. Two directed dependency graphs, with arrows pointing from a predicate to predicates or actions that depend on it. In the complete graph on the left, predicateA leads to inferredA, which leads to inferredAA and inferredAB. PredicateB leads to inferredB, which leads to inferredAB and inferredBB. InferredAB also points back to inferredB, forming a cycle. In the pruned graph on the right, inferredAA and inferredBB are removed. action1 depends on inferredA, and action2 depends on inferredB and inferredAB. The cycle is retained.
Scenario
Task actions
Reconfig. actions
Objects
Init. facts
Derived predicates
SUAVE
3
2
44
115
54
SUAVE extended
4
2
49
125
55
Navigation
1
2
537
3184
53
S1
1
2
947
8878
53
S2
1
2
150
1012
53
S3
1
2
144
1561
53
Table 2. Complexity of the planning instances. Task actions : PDDL actions encoding each task in the mission. Reconfig. actions : PDDL actions encoding architecture reconfiguration. Objects : domain constants and problem-instance objects. Init. facts : ground atoms in the initial state. Derived predicates : predicates generated by Owl2PDDL from the ontology. The scalability instances are reported at the maximum tested parameter value ( S1 : 900 nodes, S2 : 15 architectural configurations, S3 : 4 corridor types, S4 : 5 mission-level actions, and S5 : 90 nodes).
Exp.
Initial position
Water visibility
Thruster failures
Y
Min (m)
Max (m)
Period (s)
Shift (s)
Thruster
Time (s)
1
2.5+U(−0.5,0.5)
2.5
2.5
N/A
N/A
1 and 3
100+U(−100,100)
2
2.5+U(−0.5,0.5)
1.25
3.75
120
U(0,120)
N/A
N/A
3
2.5+U(−0.5,0.5)
1.25
3.75
120
U(0,120)
1 and 3
100+U(−100,100)
Table 3. Parameter configuration for the SUAVE experiments, where U(a,b) is a random uniform distribution from a to b . The initial position is relative to the Gazebo reference frame, where x is fixed at −17.0 and z at −18.5 . The time limit for each execution is set to 250 seconds.
Exp.
Managing subsystem
Search time (s)
Distance inspected (m)
Mean reaction time (s)
Mean
Std. dev.
Mean
Std. dev.
U1
U2
1
None
58.77
26.07
49.50
12.56
N/A
N/A
Random
126.71
52.90
11.65
10.08
N/A
N/A
BT
59.63
27.46
58.17
8.43
0.07
N/A
Metacontrol
59.20
25.50
58.92
8.63
1.11
N/A
ROSA
65.33
27.48
57.76
9.06
0.75
N/A
Table 4. SUAVE results. The Random managing subsystem found the pipeline in 21/30 runs (70.00%) for Experiment 1, 22/30 runs (73.33%) for Experiment 2, and 22/30 runs (73.33%) for Experiment 3. All other managing subsystems found the pipeline in every run. Search time and inspected distance report the mean and sample standard deviation only for runs that found the pipeline. U1 denotes thruster failure and U2 water visibility. Reaction times are arithmetic means of recorded events per uncertainty across all retained runs. N/A denotes an inapplicable metric or insufficient observations. The initial visibility adjustment in Experiment 1 is excluded.
Exp.
Managing subsystem
Search time (s)
Distance inspected (m)
Mean reaction time (s)
Mean
Std. dev.
Mean
Std. dev.
U1
U2
U3
1
BT
64.30
34.37
23.40
8.17
0.09
N/A
1.09
ROSA
61.10
24.85
22.41
6.95
0.81
N/A
1.95
PLANTA
71.47
37.23
17.47
8.79
0.89
N/A
1.44
2
BT
83.93
30.88
16.74
9.09
N/A
0.12
1.08
ROSA
88.70
31.50
15.40
8.96
N/A
1.57
2.03
Table 5. SUAVE extended results. All managing subsystems found the pipeline in every run. Search time and inspected distance report the mean and sample standard deviation only for runs that found the pipeline. U1 denotes thruster failure and U2 water visibility, with U3 denoting battery level. Reaction times are arithmetic means of recorded events per uncertainty across all retained runs. N/A denotes an inapplicable metric or insufficient observations. The initial visibility adjustment in Experiment 1 is excluded.
Table 6. Paired Wilcoxon signed-rank comparisons for experiment 3 and extended experiment 3. Each cell reports the unadjusted p-values for the comparison between the managing subsystem in the row and the one in the column for both the pipeline search time and pipeline distance inspected metrics. Cells with a green background indicate that the alternative hypothesis was favored, and cells with a blue background indicate that the null hypothesis cannot be rejected in favor of the alternative hypothesis. Values rounding to 0.000 are shown as <0.001 . “MC” stands for Metacontrol.
Figure 7. Comparison between PLANTA and Camara . The number is the median value, the boxes show interquartile range (Q1–Q3), the whiskers reach until ±1.5 IQR, and the circles show a sample of the observations beyond the whiskers. Two box-plot panels compare the replicated approach \CamaraApproachwith \acronym. The left panel shows planning time in seconds, with medians of 1.21 for the replicated approach and 0.40 for \acronym. The right panel shows peak memory in MiB, with medians of 461.09 and 67.39, respectively. The replicated approach has a wider spread and larger high-value outliers in both panels.
Figure 8. Grid map example with 45 retained nodes out of 50 generated nodes. White edges represent normal corridors, filled edges represent dark corridors and curved edges represent unsafe corridors. The initial location is marked with I and the goal location is marked with G . Grid-like navigation graph with 45 locations, consisting of a seven-by-seven grid with five locations removed and an additional goal location. Several connections are omitted. The initial location I is at the lower left, and the goal G is above the upper-left grid location. Edges connect neighboring locations: white straight edges denote normal corridors, filled edges denote dark corridors, and curved edges denote unsafe corridors. Filled curved edges denote corridors that are both dark and unsafe.
Experiment
Map nodes
Arch. configs
Corridor types
Actions
S1
Map size
10 – 1000 , step 10
6
2
1
S2
Arch. configurations
100
2,4,6,8,10,12,15
2
1
S3
Quality requirements
100
6
2,3,4
1
S4
Mission-level actions
100
6 – 14
2
1,2,3,4,5
S5
Combined (stress test)
10 – 100 , step 10
23
4
5
Table 7. Parameter configurations for the five scalability experiments (S1–S5). Map nodes are generated nodes. The map generator skips 10% of them, so the maximum of 1000 generated nodes in S1 and 100 in S5 corresponds to 900 and 90 retained nodes, respectively. For S4 and S5, the number of architectural configurations grows with the number of action types: each additional mission-level PDDL action introduces two further configurations.
Exp.
A*
Eager greedy
Planning time (s)
Plan length
Planning time (s)
Plan length
Mean
Std.
Mean
Std.
Mean
Std.
Mean
Std.
S1
0.92
0.02
47.20
7.74
0.95
0.03
47.20
7.74
S2
0.30
0.00
20.70
1.42
0.30
0.02
20.70
1.42
S3
0.20
0.00
20.30
2.26
0.20
0.00
20.30
2.26
S4
15.45
1.24
46.00
5.52
1.05
0.30
46.80
6.21
Table 8. Scalability experiment results at the maximum tested parameter value for each experiment ( S1 : 900 nodes, S2 : 15 architectural configurations, S3 : 4 corridor types, S4 : 5 mission-level actions, and S5 : 90 nodes).
Figure 9. Planning time in the scalability experiments S1 , S4 , and S5 . The shaded area shows the standard deviation. EG stands for the eager greedy search. Three line-plot panels show mean planning time in seconds. In S1, as map size increases from 9 to 900 retained nodes, A* and eager greedy search have similar planning times, reaching 0.92 and 0.95 seconds, respectively, and the no-adaptation baseline reaches about 0.15 seconds. In S4, as the number of mission-level action schemas increases from 1 to 5, A* planning time rises sharply to 15.45 seconds, while eager greedy search reaches 1.05 seconds and the no-adaptation baseline remains near 0.10 seconds. In S5, the combined stress test, as map size increases from 9 to 90 retained nodes, A* planning time rises to 188.96 seconds, whereas eager greedy search reaches 13.73 seconds. Shaded bands indicate standard deviations.
Figure 10. S1 results: Planning time and plan length as map size increases. The shaded area shows the standard deviation. EG stands for the eager greedy search. Two line-plot panels show mean planning time in seconds and mean plan length in action counts against map size, from 9 to 900 retained nodes. A* and eager greedy search have similar planning times, reaching 0.92 and 0.95 seconds, respectively, at 900 nodes, and overlapping plan-length curves, reaching 47.20 actions. The left panel also shows the faster no-adaptation baseline. Shaded bands indicate standard deviations.
Figure 11. S2 results: Planning time and plan length as architectural configurations increase. The shaded area shows the standard deviation. EG stands for the eager greedy search. Two line-plot panels show mean planning time in seconds and mean plan length in action counts as the number of architectural configurations increases from 2 to 15. Planning time changes modestly, reaching 0.30 seconds for both A* and eager greedy search. The plan-length curves overlap, reaching 20.70 actions at 15 configurations. Shaded bands indicate standard deviations.
Figure 12. S3 results: Planning time and plan length as corridor types increase. The shaded area shows the standard deviation. EG stands for the eager greedy search. Two line-plot panels show mean planning time in seconds and mean plan length in action counts as the number of corridor types increases from 2 to 4. A* and eager greedy search have nearly identical planning times, increasing from about 0.15 to 0.20 seconds. Their plan-length curves overlap, with means of 19.1 actions at two corridor types and 20.3 at four. Shaded bands indicate standard deviations.
Figure 13. S4 results: Planning time and plan length as mission-level actions increase. The shaded area shows the standard deviation. EG stands for the eager greedy search. Two line-plot panels show mean planning time in seconds and mean plan length in action counts as the number of mission-level action schemas increases from 1 to 5. A* planning time rises sharply to 15.45 seconds, while eager greedy search reaches 1.05 seconds and the no-adaptation baseline remains near 0.10 seconds. Plan lengths increase for both search strategies and remain similar, reaching 46.00 and 46.80 actions, respectively. Shaded bands indicate standard deviations.
Figure 14. S5 results: Planning time and plan length as map size increases. The shaded area shows the standard deviation. EG stands for the eager greedy search. Two line-plot panels show mean planning time and mean plan length against map size, from 9 to 90 retained nodes, in the combined stress test. A* planning time rises to 188.96 seconds, whereas eager greedy search reaches 13.73 seconds. Plan lengths increase with map size, reaching 48.10 actions for A* and 51.60 for eager greedy search. Shaded bands indicate standard deviations.
In many robotic manipulation tasks, the robot repeatedly solves motion-planning problems that differ mainly in the location of the goal object and its associated obstacle, while the surrounding workspace remains fixed. Prior works have shown that leveraging experience and offline computation can accelerate repeated planning queries, but they lack guarantees of covering the continuous task space and require storing large libraries of solutions. In this work, we present COAD, a framework that provides constant-time planning over a continuous goal-parameterized task space. COAD discretizes the continuous task space into finitely many Task Coverage Regions. Instead of planning and storing solutions for every region offline, it constructs a compressed library by only solving representative root problems. Other problems are handled through fast adaptation from these root solutions. At query time, the system retrieves a root motion in constant time and adapts it to the desired goal using lightweight adaptation modules such as linear interpolation, Dynamic Movement Primitives, or simple trajectory optimization. We evaluate the framework on various manipulators and environments in simulation and the real world, showing that COAD achieves substantial compression of the motion library while maintaining high success rates and sub-millisecond-level queries, outperforming baseline methods in both efficiency and path quality. The source code is available at https://github.com/elpis-lab/CoAd.
Adil Shiyas, Zhuoyun Zhong, Constantinos Chamzas
Department of Robotics Engineering, Worcester Polytechnic Institute (WPI), Worcester, MA 01609, USA
Task planning for robots in off-site construction prefabrication is primarily manual, leading to planning inefficiencies and reluctance to adopt robotic solutions. Existing approaches provide limited support for scalable, modular automated task planning and localized runtime replanning. This paper proposes APTree, a model-driven framework that integrates automated task planning with modular task execution using behavior trees as the execution model. APTree repeatedly applies and coordinates heterogeneous symbolic planners to automatically generate and compose modular task plans at runtime. To support this approach, we develop a domain-specific language for the formal specification and validation of task plans, together with a compatible execution engine. We evaluate APTree on the real-world livMatS Biomimetic Shell pavilion, where it models the construction robotic fabrication domain, detects injected specification and planning errors at design time and runtime, and automatically generates large-scale task plans by coordinating three symbolic planners across numerous planning modules. A machine-checked formal analysis further shows that composed execution and local replanning are safe under explicitly stated assumptions, including planner soundness and non-interference of generated plans, providing practitioners with explicit conditions for dependable plan composition. APTree thus enables construction practitioners to leverage automated task planning without requiring detailed knowledge of symbolic planning techniques or APTree's internal implementation.
Shermin Sherkat, Thomas Wortmann, Andreas Wortmann
Institute for Computational Design and Construction, Department for Computing in Architecture (ICD/CA), University of Stuttgart, Keplerstraße 11, Stuttgart, 70174, Germany · Institute for Control Engineering of Machine Tools and Manufacturing Units (ISW), University of Stuttgart, Seidenstraße 36, Stuttgart, 70174, Germany
Anticipating and adapting to failures is a key capability robots need to collaborate effectively with humans in complex domains. This continues to be a challenge despite the impressive performance of state of the art AI planning systems and Large Language Models (LLMs) because of the uncertainty associated with the tasks and their outcomes. Toward addressing this challenge, we present a hybrid framework that integrates the generic prediction capabilities of an LLM with the probabilistic sequential decision-making capability of Relational Dynamic Influence Diagram Language. For any given task, the robot reasons about the task and the capabilities of the human attempting to complete it; predicts potential failures due to lack of ability (in the human) or lack of relevant domain objects; and executes actions to prevent such failures or recover from them. Experimental evaluation in the VirtualHome 3D simulation environment demonstrates substantial improvement in performance compared with state of the art baselines.
Nabanita Dash, Ayush Kaura, Shivam Singh +4
Robotics Research Center, IIIT Hyderabad, India · TCS Research, Tata Consultancy Services, India · School of Informatics, University of Edinburgh, UK