Fig. 1. An overview of our approach. Since creating large numbers of realistic object models is challenging, we train our deep autoregressive model architecture on millions of unrealistic procedurally generated objects (indicated in blue above) and billions of unique grasp attempts. At test time, our model generalizes to realistic objects from the YCB dataset (indicated in green above) [4] with 92% success rate.Abstract-Deep learning-based robotic grasping has made significant progress thanks to algorithmic improvements and increased data availability. However, state-of-the-art models are often trained on as few as hundreds or thousands of unique object instances, and as a result generalization can be a challenge.In this work, we explore a novel data generation pipeline for training a deep neural network to perform grasp planning that applies the idea of domain randomization to object synthesis. We generate millions of unique, unrealistic procedurally generated objects, and train a deep neural network to perform grasp planning on these objects.Since the distribution of successful grasps for a given object can be highly multimodal, we propose an autoregressive grasp planning model that maps sensor inputs of a scene to a probability distribution over possible grasps. This model allows us to sample grasps efficiently at test time (or avoid sampling entirely). We evaluate our model architecture and data generation pipeline in simulation and the real world. We find we can achieve a >90% success rate on previously unseen realistic objects at test time in simulation despite having only been trained on random objects. We also demonstrate an 80% success rate on real-world grasp attempts despite having only been trained on random simulated objects.

No abstract

Crowdsourcing is an effective tool to solve hard tasks. By bringing 100,000s of people to work on simple tasks that only humans can do, we can go far beyond traditional models of data analysis and machine learning. As technologies and processes mature, crowdsourcing is becoming mainstream. It powers many leading Internet companies and a wide variety of novel projects: from content moderation and business listing verification to real-time SMS translation for disaster response. However, quality assurance can be a major challenge. In this paper CrowdFlower presents various crowdsourcing applications, from business to ethics, to money and survival, all of which showcase the power of labor-on-demand, otherwise known as the human cloud.

Crowdsourced crisis response harnesses distributed networks of humans in combination with information and communication technology (ICT) to create scalable, flexible and rapid communication systems that promote well-being, survival, and recovery during the acute phase of an emergency. In this paper, we analyze a recent experience in which CrowdFlower conducted crowdsourced translation, categorization and geo-tagging for SMS-based reporting as part of Mission 4636 after a 7.0 magnitude earthquake struck Haiti on January 12, 2010. We discuss CrowdFlower's approach to this task, lessons learned from the experience, and opportunities to generalize the techniques and technologies involved for other ICT for development (ICTD) applications. We find that CrowdFlower's most significant contribution to Mission 4636 and to the broader field of crowdsourced crisis relief lies in the flexible, scalable nature of the pool of earthquake survivors, volunteers, workers, and machines that the organization engaged during the emergency response efforts.

Crowdsourcing is an effective tool to solve hard tasks. By bringing 100,000s of people to work on simple tasks that only humans can do, we can go far beyond traditional models of data analysis and machine learning. As technologies and processes mature, crowdsourcing is becoming mainstream. It powers many leading Internet companies and a wide variety of novel projects: from content moderation and business listing verification to real-time SMS translation for disaster response. However, quality assurance can be a major challenge. In this paper CrowdFlower presents various crowdsourcing applications, from business to ethics, to money and survival, all of which showcase the power of labor-on-demand, otherwise known as the human cloud.