Stefano Carpin
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Research Interests
Stefano Carpin has a total of 75 co-authors. Here are the top co-authors by number of shared publications:
Andreas Kolling
School of Information Sciences, University of Pittsburgh, PA, USA
3 shared publicationsBenjamin Balaguer
School of Engineering, University of California, Merced, USA
3 shared publications- Marco Pavone
Stanford University
3 shared publications - Arnoud Visser
University of Amsterdam
2 shared publications - Brian M. Sadler
Oden Institute for Computational Engineering and Sciences, The University of Texas at Austin, Austin, TX 78712, USA
2 shared publications Gorkem Erinc
School of Engineering, University of California, Merced (USA)
2 shared publications
Latest Publications
- 9 October 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Automation Science and Engineering
- Vol. 22 (15455955) , 7855-7869
Solving Stochastic Orienteering Problems With Chance Constraints Using Monte Carlo Tree Search
We present a new Monte Carlo Tree Search (MCTS) algorithm to solve the stochastic orienteering problem with chance constraints, i.e., a version of the problem where travel costs are random, and one is assigned a bound on the tolerable probability of exceeding the budget. The algorithm we present is online and anytime, i.e., it alternates planning and execution, and the quality of the solution it produces increases as the allowed computational time increases. Differently from most former MCTS algorithms, for each action available in a state the algorithm maintains estimates of both its value and the probability that its execution will eventually result in a violation of the chance constraint. Then, at action selection time, our proposed solution prunes away trajectories that are estimated to violate the failure probability. Extensive simulation results show that this approach can quickly produce high-quality solutions and is competitive with the optimal but time-consuming solution. Note to Practitioners —In many practical scenarios one is faced with multiobjective sequential decision making problems that can be solved through constrained optimization. If some of the parameters are known with uncertainty, the event “violating one of the constraints” becomes a random variable whose probability should be bound. As an application of this general problem formulation, in this paper we consider stochastic orienteering, a problem that finds applications when a robot is tasked with performing multiple tasks of varying utility while being subject to a bound on the traveled distance. Many problems in logistics, precision agriculture, and environmental monitoring, just to name a few, can be cast as instances of this optimization problem.
- 9 October 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Automation Science and Engineering
- Vol. 22 (15455955) , 7855-7869
Solving Stochastic Orienteering Problems With Chance Constraints Using Monte Carlo Tree Search
We present a new Monte Carlo Tree Search (MCTS) algorithm to solve the stochastic orienteering problem with chance constraints, i.e., a version of the problem where travel costs are random, and one is assigned a bound on the tolerable probability of exceeding the budget. The algorithm we present is online and anytime, i.e., it alternates planning and execution, and the quality of the solution it produces increases as the allowed computational time increases. Differently from most former MCTS algorithms, for each action available in a state the algorithm maintains estimates of both its value and the probability that its execution will eventually result in a violation of the chance constraint. Then, at action selection time, our proposed solution prunes away trajectories that are estimated to violate the failure probability. Extensive simulation results show that this approach can quickly produce high-quality solutions and is competitive with the optimal but time-consuming solution. Note to Practitioners —In many practical scenarios one is faced with multiobjective sequential decision making problems that can be solved through constrained optimization. If some of the parameters are known with uncertainty, the event “violating one of the constraints” becomes a random variable whose probability should be bound. As an application of this general problem formulation, in this paper we consider stochastic orienteering, a problem that finds applications when a robot is tasked with performing multiple tasks of varying utility while being subject to a bound on the traveled distance. Many problems in logistics, precision agriculture, and environmental monitoring, just to name a few, can be cast as instances of this optimization problem.
- 1 February 2024
- Published by Springer Nature
- p. 466-478
Distributed Estimation of Scalar Fields with Implicit Coordination
- 1 February 2024
- Published by Springer Nature
- p. 466-478
Distributed Estimation of Scalar Fields with Implicit Coordination
- Open Access
- 26 October 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics & Automation Magazine
- Vol. 30 (4) , 52-67
Robotic Assessment of a Crop’s Need for Watering: Automating a Time-Consuming Task to Support Sustainable Agriculture
Agricultural robotics and automation technology plays an increasingly critical role across several crop production stages to improve sustainability (e.g., water use optimization). Robotics has been applied for remote and proximal sensing as well as physical sampling. The procedure of physical sampling and follow-on analysis of plant specimens, such as leaves or shoots, which is the focus of this work, often constitutes the only accurate way to measure some essential parameters that affect crop production, such as stem water potential (SWP), which helps determine a crop’s need for watering.
- 26 October 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics & Automation Magazine
- Vol. 30 (4) , 52-67
Open AccessRobotic Assessment of a Crop’s Need for Watering: Automating a Time-Consuming Task to Support Sustainable Agriculture
Agricultural robotics and automation technology plays an increasingly critical role across several crop production stages to improve sustainability (e.g., water use optimization). Robotics has been applied for remote and proximal sensing as well as physical sampling. The procedure of physical sampling and follow-on analysis of plant specimens, such as leaves or shoots, which is the focus of this work, often constitutes the only accurate way to measure some essential parameters that affect crop production, such as stem water potential (SWP), which helps determine a crop’s need for watering.
- 1 September 2022
- Published by Elsevier in Robotics and Autonomous Systems
- Vol. 157, 104266
Trust as a metric for auction-based task assignment in a cooperative team of robots with heterogeneous capabilities
- 1 September 2022
- Published by Elsevier in Robotics and Autonomous Systems
- Vol. 157, 104266
Trust as a metric for auction-based task assignment in a cooperative team of robots with heterogeneous capabilities
- 17 January 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Automation Science and Engineering
- Vol. 19 (3) , 1352-1364
Simulating Polyculture Farming to Learn Automation Policies for Plant Diversity and Precision Irrigation
Polyculture farming, where multiple crop species are grown simultaneously, has potential to reduce pesticide and water usage while improving the utilization of soil nutrients. However, it is much harder to automate polyculture than monoculture. To facilitate research, we present AlphaGardenSim, a fast, first order, open-access polyculture farming simulator with single plant growth and irrigation models tuned using real world measurements. AlphaGardenSim can be used for policy learning as it simulates inter-plant dynamics, including light and water competition between plants in close proximity and approximates growth in a real greenhouse garden at 25, $000\times $ the speed of natural growth. This paper extends earlier work with a new action space that includes planting, which dynamically finds new seed locations that increases resources utilization, and an adaptive sampling technique to reduce the number of actions taken at each timestep without affecting performance. We also evaluate other automation policies using a novel metric that combines plant diversity and canopy coverage. Code and supplementary material can be found at https://github.com/BerkeleyAutomation/AlphaGarden . Note to Practitioners —Monoculture farming is often characterized by heavy agrichemical inputs, such as chemical fertilizers and pesticides, and increased vulnerability to disease and pestilence. This paper is motivated by the lack of long-term sustainability of industrial agriculture, and its implications for human food security. Although polyculture is a sustainable alternative to monoculture farming, it requires more human labor and is more challenging to automate. In this paper we propose a fast, first order simulator that simulates the growth of plants in a polyculture setting. Simulation experiments suggest that the simulator can be used to learn a planting, watering and pruning plan a robot can follow to produce maximal yield from a diverse set of plants with limited irrigation, however it has not yet been tested on a physical garden. In future research we will develop a fully automated controller that will operate planting, irrigation and pruning tools in a physical garden over multiple plant growth cycles.
- 17 January 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Automation Science and Engineering
- Vol. 19 (3) , 1352-1364
Simulating Polyculture Farming to Learn Automation Policies for Plant Diversity and Precision Irrigation
Polyculture farming, where multiple crop species are grown simultaneously, has potential to reduce pesticide and water usage while improving the utilization of soil nutrients. However, it is much harder to automate polyculture than monoculture. To facilitate research, we present AlphaGardenSim, a fast, first order, open-access polyculture farming simulator with single plant growth and irrigation models tuned using real world measurements. AlphaGardenSim can be used for policy learning as it simulates inter-plant dynamics, including light and water competition between plants in close proximity and approximates growth in a real greenhouse garden at 25, $000\times $ the speed of natural growth. This paper extends earlier work with a new action space that includes planting, which dynamically finds new seed locations that increases resources utilization, and an adaptive sampling technique to reduce the number of actions taken at each timestep without affecting performance. We also evaluate other automation policies using a novel metric that combines plant diversity and canopy coverage. Code and supplementary material can be found at https://github.com/BerkeleyAutomation/AlphaGarden . Note to Practitioners —Monoculture farming is often characterized by heavy agrichemical inputs, such as chemical fertilizers and pesticides, and increased vulnerability to disease and pestilence. This paper is motivated by the lack of long-term sustainability of industrial agriculture, and its implications for human food security. Although polyculture is a sustainable alternative to monoculture farming, it requires more human labor and is more challenging to automate. In this paper we propose a fast, first order simulator that simulates the growth of plants in a polyculture setting. Simulation experiments suggest that the simulator can be used to learn a planting, watering and pruning plan a robot can follow to produce maximal yield from a diverse set of plants with limited irrigation, however it has not yet been tested on a physical garden. In future research we will develop a fully automated controller that will operate planting, irrigation and pruning tools in a physical garden over multiple plant growth cycles.
- 10 June 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Robotics
- Vol. 38 (1) , 433-447
Speeding up Routing Schedules on Aisle Graphs With Single Access
In this article, we study the orienteering aisle-graph single-access problem (OASP), a variant of the orienteering problem for a robot moving in a so-called single-access aisle graph , i.e., a graph consisting of a set of rows that can be accessed from one side only. Aisle graphs model, among others, vineyards or warehouses. Each aisle-graph vertex is associated with a reward that a robot obtains when it visits the vertex itself. As the energy of the robot is limited, only a subset of vertices can be visited with a fully charged battery. The objective is to maximize the total reward collected by the robot with a battery charge. We first propose an optimal algorithm that solves the OASP in $\mathcal {O}(m^2n^2)$ time for aisle graphs with a single access consisting of $m$ rows, each with $n$ vertices. With the goal of designing faster solutions, we propose four greedy suboptimal algorithms that run in at most $\mathcal {O}(mn\ (m + n))$ time. For two of them, we guarantee an approximation ratio of $\frac{1}{2}(1 - \frac{1}{e})$ , where $e$ is the base of the natural logarithm, on the total reward by exploiting the well-known submodularity property. Experimentally, we show that these algorithms collect more than 80% of the optimal reward.
- 10 June 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Robotics
- Vol. 38 (1) , 433-447
Speeding up Routing Schedules on Aisle Graphs With Single Access
In this article, we study the orienteering aisle-graph single-access problem (OASP), a variant of the orienteering problem for a robot moving in a so-called single-access aisle graph , i.e., a graph consisting of a set of rows that can be accessed from one side only. Aisle graphs model, among others, vineyards or warehouses. Each aisle-graph vertex is associated with a reward that a robot obtains when it visits the vertex itself. As the energy of the robot is limited, only a subset of vertices can be visited with a fully charged battery. The objective is to maximize the total reward collected by the robot with a battery charge. We first propose an optimal algorithm that solves the OASP in $\mathcal {O}(m^2n^2)$ time for aisle graphs with a single access consisting of $m$ rows, each with $n$ vertices. With the goal of designing faster solutions, we propose four greedy suboptimal algorithms that run in at most $\mathcal {O}(mn\ (m + n))$ time. For two of them, we guarantee an approximation ratio of $\frac{1}{2}(1 - \frac{1}{e})$ , where $e$ is the base of the natural logarithm, on the total reward by exploiting the well-known submodularity property. Experimentally, we show that these algorithms collect more than 80% of the optimal reward.
- 24 March 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 6 (2) , 4185-4192
An Adaptive Method for the Stochastic Orienteering Problem
We consider the NP-hard Stochastic Orienteering Problem, where the goal is to navigate between start and end vertices in a graph, maximizing the sum of rewards for visited vertices while obeying a travel budget over edges with stochastic cost within a given probability of failure. Previously, we solved this by finding an initial path using a deterministic orienteering solver and transformed it into a path policy using a Constrained Markov Decision Process that can skip vertices based on arrival time. In this work we augment our technique, creating a path tree which branches at vertex-time states with high probability of being skipped, allowing for new sequences of vertices in the resulting policy. We demonstrate that this adaptive path method collects significantly more reward in expectation, even when the number of branches is limited to control computation time.
- 24 March 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 6 (2) , 4185-4192
An Adaptive Method for the Stochastic Orienteering Problem
We consider the NP-hard Stochastic Orienteering Problem, where the goal is to navigate between start and end vertices in a graph, maximizing the sum of rewards for visited vertices while obeying a travel budget over edges with stochastic cost within a given probability of failure. Previously, we solved this by finding an initial path using a deterministic orienteering solver and transformed it into a path policy using a Constrained Markov Decision Process that can skip vertices based on arrival time. In this work we augment our technique, creating a path tree which branches at vertex-time states with high probability of being skipped, allowing for new sequences of vertices in the resulting policy. We demonstrate that this adaptive path method collects significantly more reward in expectation, even when the number of branches is limited to control computation time.
- Open Access
- 25 February 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 6 (2) , 3287-3294
Task Planning on Stochastic Aisle Graphs for Precision Agriculture
This work addresses task planning under uncertainty for precision agriculture applications whereby task costs are uncertain and the gain of completing a task is proportional to resource consumption (such as water consumption in precision irrigation). The goal is to complete all tasks while prioritizing those that are more urgent, and subject to diverse budget thresholds and stochastic costs for tasks. To describe agriculture-related environments that incorporate stochastic costs to complete tasks, a new Stochastic-Vertex-Cost Aisle Graph (SAG) is introduced. Then, a task allocation algorithm, termed Next-Best-Action Planning (NBA-P), is proposed. NBA-P utilizes the underlying structure enabled by SAG, and tackles the task planning problem by simultaneously determining the optimal tasks to perform and an optimal time to exit (i.e. return to a base station), at run-time. The proposed approach is tested with both simulated data and real-world experimental datasets collected in a commercial vineyard, in both single- and multi-robot scenarios. In all cases, NBA-P outperforms other evaluated methods in terms of return per visited vertex, wasted resources resulting from aborted tasks (i.e. when a budget threshold is exceeded), and total visited vertices.
- 25 February 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 6 (2) , 3287-3294
Open AccessTask Planning on Stochastic Aisle Graphs for Precision Agriculture
This work addresses task planning under uncertainty for precision agriculture applications whereby task costs are uncertain and the gain of completing a task is proportional to resource consumption (such as water consumption in precision irrigation). The goal is to complete all tasks while prioritizing those that are more urgent, and subject to diverse budget thresholds and stochastic costs for tasks. To describe agriculture-related environments that incorporate stochastic costs to complete tasks, a new Stochastic-Vertex-Cost Aisle Graph (SAG) is introduced. Then, a task allocation algorithm, termed Next-Best-Action Planning (NBA-P), is proposed. NBA-P utilizes the underlying structure enabled by SAG, and tackles the task planning problem by simultaneously determining the optimal tasks to perform and an optimal time to exit (i.e. return to a base station), at run-time. The proposed approach is tested with both simulated data and real-world experimental datasets collected in a commercial vineyard, in both single- and multi-robot scenarios. In all cases, NBA-P outperforms other evaluated methods in terms of return per visited vertex, wasted resources resulting from aborted tasks (i.e. when a budget threshold is exceeded), and total visited vertices.
- Open Access
- 7 October 2020
- Published by MDPI AG in Remote Sensing
- Vol. 12 (19) , 3251
sUAS Remote Sensing of Vineyard Evapotranspiration Quantifies Spatiotemporal Uncertainty in Satellite-Borne ET Estimates
Small Unmanned Aerial Systems (sUAS) show promise in being able to collect high resolution spatiotemporal data over small extents. Use of such remote sensing platforms also show promise for quantifying uncertainty in more ubiquitous Earth Observation System (EOS) data, such as evapotranspiration and consumptive use of water in agricultural systems. This study compares measurements of evapotranspiration (ET) from a commercial vineyard in California using data collected from sUAS and EOS sources for 10 events over a growing season using multiple ET estimation methods. Results indicate that sUAS ET estimates that include non-canopy pixels are generally lower on average than EOS methods by >0.5 mm day. sUAS ET estimates that mask out non-canopy pixels are generally higher than EOS methods by <0.5 mm day. Masked sUAS ET estimates are less variable than unmasked sUAS and EOS ET estimates. This study indicates that limited deployment of sUAS can provide important estimates of uncertainty in EOS ET estimations for larger areas and to also improve irrigation management at a local scale.- 7 October 2020
- Published by MDPI AG in Remote Sensing
- Vol. 12 (19) , 3251
Open AccesssUAS Remote Sensing of Vineyard Evapotranspiration Quantifies Spatiotemporal Uncertainty in Satellite-Borne ET Estimates
Small Unmanned Aerial Systems (sUAS) show promise in being able to collect high resolution spatiotemporal data over small extents. Use of such remote sensing platforms also show promise for quantifying uncertainty in more ubiquitous Earth Observation System (EOS) data, such as evapotranspiration and consumptive use of water in agricultural systems. This study compares measurements of evapotranspiration (ET) from a commercial vineyard in California using data collected from sUAS and EOS sources for 10 events over a growing season using multiple ET estimation methods. Results indicate that sUAS ET estimates that include non-canopy pixels are generally lower on average than EOS methods by >0.5 mm day. sUAS ET estimates that mask out non-canopy pixels are generally higher than EOS methods by <0.5 mm day. Masked sUAS ET estimates are less variable than unmasked sUAS and EOS ET estimates. This study indicates that limited deployment of sUAS can provide important estimates of uncertainty in EOS ET estimations for larger areas and to also improve irrigation management at a local scale.