Fleet Management of Autonomous Agricultural Robots with Human Awareness

自主农业机器人的车队管理是完全自主农业实践的基础，因为它解决了在处理同质或异质机器人车队时解决冲突、资源和效率优化的问题。这篇博士论文的摘要是将人类行为模型与自主车队的导航和任务分配系统相结合，以更好地规划路径，避免机器人-机器人和人类-机器人之间可能发生的冲突和死锁情况。研究将集中在邻近学，人类行为建模使用决策预测和神经科学方法，路径规划，车队协调和人类机器人伦理。研究中提出的想法计划首先使用合适的计算机模拟软件进行测试，并根据需要使用人类导航和行为模式的真实世界数据。模拟环境应该是一个农业环境，具有像Thorvald （Saga Robotics）这样的自主农业机器人的现实表现。作为更高级研究的一部分，通过计算机模拟获得的成功想法需要通过在人类参与的农业环境中植入类似的真实机器人来进行测试。所得结果可用于评估、修改和改进所提出的想法。从现实世界中收集人类导航数据将在项目实施阶段的几个月后出现。由于正在发生的COVID大流行，其可行性存在风险，作为一项风险缓解战略，如果条件没有改善到足以使这种数据收集成为可能，则用于计算机模拟的人类行为模型的现有科学数据应使用应在数据收集再次成为可能时被真实数据所取代。

Fleet Management of Autonomous Agricultural Robots is fundamental to fully autonomous farming practices as it addresses the issues of conflict resolution, resource and efficiency optimization when dealing with a fleet of homogenous or heterogenous robots. The abstract of the proposal for this PhD would be integrating human behaviour models with the navigation and task allocation system of autonomous fleets for better path planning evading possible conflicts and deadlock situations between robot-robot and human-robot(s). The study will focus on the proxemics, human behaviour modelling using decision prediction and neuroscientific methods, path planning, fleet coordination and human robot ethics. The ideas proposed in the study are planned to be first tested out using a suitable computer simulation software with the help of real world data of human navigation and behavioural patterns as needed. The simulation environment shall be an agricultural one with realistic representations of autonomous agri-robots like Thorvald (by Saga Robotics). As a part of more advanced research the successful ideas obtained through computer simulation shall be tested through implantation on similar real robots in an agricultural environment with human participation. The results obtained shall be used for evaluation, revision and improvisation of the proposed ideas. The collection of human navigation data from the real world shall come up after a few months during the implementation phase of the project. There is a risk of its feasibility due to the ongoing COVID pandemic, as a risk mitigation strategy, if the conditions do not improve well enough to enable this data collection possible, existing scientific data of human behavioural models shall be used for the computer simulations and shall be replaced with real data when data collection becomes possible again.