摘要：【中文】本发明公开一种基于传感信息融合的无人驾驶避障方法及系统，本系统用于实现本方法，本方法包括获取在时间段[t₁,t₂]间无人车行驶距离；获取无人车在时间t₁的行驶距离和在时间t₂的距离计算在时间段[t₁,t₂]间无人车行驶距离与无人车至障碍物的距离差值的比例值；将该比例值与预定范围进行比较，若该比例值在预定范围内，则将无人车行驶距离与无人车至障碍物的距离差值加权平均以获取无人车至障碍物的修正距离若该比例值不在预定范围内，则视无人车行驶距离为无人车至障碍物的修正距离根据无人车至障碍物的修正距离控制无人车的避障动作。本发明减少由障碍物深度信息的噪声污染和环境变化导致无人车至障碍物距离的测量误差，提高无人车安全性。 【EN】The invention discloses an unmanned obstacle avoidance method and system based on sensing information fusion₁,t₂]The running distance of the unmanned vehicle is reduced; obtaining the time t of the unmanned vehicle₁Distance traveledAnd at time t₂Is a distance ofCalculating over a time period t₁,t₂]Distance between unmanned vehicles and unmanned vehiclesA proportional value of a distance difference of the obstacle; comparing the ratio value with a preset range, and if the ratio value is in the preset range, carrying out weighted average on the driving distance of the unmanned vehicle and the distance difference between the unmanned vehicle and the obstacle to obtain the corrected distance between the unmanned vehicle and the obstacleIf the ratio value is not in the preset range, the distance between the unmanned vehicle and the obstacle is considered as the correction distanceAccording to the corrected distance from the unmanned vehicle to the obstacleAnd controlling the obstacle avoidance action of the unmanned vehicle. The invention reduces the measurement error of the distance between the unmanned vehicle and the obstacle caused by the noise pollution of the depth information of the obstacle and the environmental change, and improves the safety of the unmanned vehicle.

摘要：【中文】本发明属于计算机技术领域，具体涉及基于物理距离和语义距离的双向LSTM模型的构建方法，包括以下步骤：a、进行数据预处理；b、对LSTM模型中的单词进行向量化处理并对其物理特征和语义特征进行设置；c、构建双向LSTM模型。与现有技术相比，本发明可以有效地在LSTM网络的输入上增加物理特征和语义特征，在特征选择上，不仅仅使用词向量作为特征，更实将位置特征添加进来，将模型的训练集从句子级别细化到了实体对级别；模型构建上，也考虑到了语义上的句法以来，既充分考虑到了两个实体之间的最直接语义特征，降低上下文长度，又考虑到了非最短路径上的词语对分类结果的影响，有效提高了准确率。 【EN】The invention belongs to the technical field of computers, and particularly relates to a method for constructing a bidirectional LSTM model based on physical distance and semantic distance, which comprises the following steps: a. carrying out data preprocessing; b. vectorizing the words in the LSTM model and setting the physical characteristics and semantic characteristics of the words; c. and constructing a bidirectional LSTM model. Compared with the prior art, the method can effectively add physical characteristics and semantic characteristics on the input of the LSTM network, not only uses word vectors as characteristics on characteristic selection, but also more practically adds position characteristics, and refines the training set of the model from sentence level to entity pair level; in the aspect of model construction, the syntax on semantics is also considered, the most direct semantic features between two entities are fully considered, the context length is reduced, the influence of words on the classification result on the non-shortest path is considered, and the accuracy is effectively improved.

摘要：【中文】本发明公开了一种低层装配式建筑，包括全预制基础连墙、全预制剪力墙、连系梁、非承重外墙、内隔墙和屋面板，全预制基础连墙和全预制剪力墙通过金属连接件连接成为横向剪力墙，连系梁连接在相邻的横向剪力墙之间，屋面板与全预制剪力墙连接，相邻的横向剪力墙之间通过连系梁、非承重外墙、内隔墙和屋面板连接起来构成整体的空间结构。相应地，本发明还公开了实施前述所述一种低层装配式建筑的建造方法。本发明针对主流装配式结构存在的不足进行改进，具备多项技术优势和成本优势，适合国内或海外低层远距离运输项目。 【EN】The invention discloses a low-rise fabricated building which comprises a fully-prefabricated foundation connecting wall, a fully-prefabricated shear wall, connecting beams, non-bearing outer walls, inner partition walls and a roof panel. Correspondingly, the invention also discloses a construction method for implementing the low-rise fabricated building. The invention improves the defects of the mainstream assembled structure, has multiple technical advantages and cost advantages, and is suitable for domestic or overseas low-rise long-distance transportation projects.